AI Writing Tools

Explore the best AI Writing Tools — independent reviews, comparisons, pricing and step-by-step how-to guides, curated by Aizhi.

  • Trazzler

    Trazzler

    Trazzler is a travel destination app that specializes in unique and local destinations. The initial concept was developed by Adam Rugel and Biz Stone in 2006 at Twitter's original offices under the name "71 miles". More than 10,000 writers and photographers have contributed and more than $350,000 in freelance contracts have been issued as a result of Trazzeler's weekly writing and photography contests. Investors in the company include SV Angel, AOL Founder Steve Case, and the Twitter founders, Evan Williams, Jack Dorsey, and Biz Stone. The company's partners are the City of Chicago, Hawaii Tourism Authority, Fairmont Hotels & Resorts, Salon.com, and Air New Zealand. Trazzler is designed for use on the iOS, Android, and Facebook.

    Read more →
  • Shyster (expert system)

    Shyster (expert system)

    SHYSTER is a legal expert system developed at the Australian National University in Canberra in 1993. It was written as the doctoral dissertation of James Popple under the supervision of Robin Stanton, Roger Clarke, Peter Drahos, and Malcolm Newey. A full technical report of the expert system, and a book further detailing its development and testing have also been published. SHYSTER emphasises its pragmatic approach, and posits that a legal expert system need not be based upon a complex model of legal reasoning in order to produce useful advice. Although SHYSTER attempts to model the way in which lawyers argue with cases, it does not attempt to model the way in which lawyers decide which cases to use in those arguments. SHYSTER is of a general design, permitting its operation in different legal domains. It was designed to provide advice in areas of case law that have been specified by a legal expert using a bespoke specification language. Its knowledge of the law is acquired, and represented, as information about cases. It produces its advice by examining, and arguing about, the similarities and differences between cases. It derives its name from Shyster: a slang word for someone who acts in a disreputable, unethical, or unscrupulous way, especially in the practice of law and politics. == Methods == SHYSTER is a specific example of a general category of legal expert systems, broadly defined as systems that make use of artificial intelligence (AI) techniques to solve legal problems. Legal AI systems can be divided into two categories: legal retrieval systems and legal analysis systems. SHYSTER belongs to the latter category of legal analysis systems. Legal analysis systems can be further subdivided into two categories: judgment machines and legal expert systems. SHYSTER again belongs to the latter category of legal expert systems. A legal expert system, as Popple uses the term, is a system capable of performing at a level expected of a lawyer: "AI systems which merely assist a lawyer in coming to legal conclusions or preparing legal arguments are not here considered to be legal expert systems; a legal expert system must exhibit some legal expertise itself." Designed to operate in more than one legal domain, and be of specific use to the common law of Australia, SHYSTER accounts for statute law, case law, and the doctrine of precedent in areas of private law. Whilst it accommodates statute law, it is primarily a case-based system, in contradistinction to rule-based systems like MYCIN. More specifically, it was designed in a manner enabling it to be linked with a rule-based system to form a hybrid system. Although case-based reasoning possesses an advantage over rule-based systems by the elimination of complex semantic networks, it suffers from intractable theoretical obstacles: without some further theory it cannot be predicted what features of a case will turn out to be relevant. Users of SHYSTER therefore require some legal expertise. Richard Susskind argues that "jurisprudence can and ought to supply the models of law and legal reasoning that are required for computerized [sic] implementation in the process of building all expert systems in law". Popple, however, believes jurisprudence is of limited value to developers of legal expert systems. He posits that a lawyer must have a model of the law (maybe unarticulated) which includes assumptions about the nature of law and legal reasoning, but that model need not rest on basic philosophical foundations. It may be a pragmatic model, developed through experience within the legal system. Many lawyers perform their work with little or no jurisprudential knowledge, and there is no evidence to suggest that they are worse, or better, at their jobs than lawyers well-versed in jurisprudence. The fact that many lawyers have mastered the process of legal reasoning, without having been immersed in jurisprudence, suggests that it may indeed be possible to develop legal expert systems of good quality without jurisprudential insight. As a pragmatic legal expert system SHYSTER is the embodiment of this belief. A further example of SHYSTER’s pragmatism is its simple knowledge representation structure. This structure was designed to facilitate specification of different areas of case law using a specification language. Areas of case law are specified in terms of the cases and attributes of importance in those areas. SHYSTER weights its attributes and checks for dependence between them. In order to choose cases upon which to construct its opinions, SHYSTER calculates distances between cases and uses these distances to determine which of the leading cases are nearest to the instant case. To this end SHYSTER can be seen to adopt and expand upon nearest neighbor search methods used in pattern recognition. These nearest cases are used to produce an argument (based on similarities and differences between the cases) about the likely outcome in the instant case. This argument relies on the doctrine of precedent; it assumes that the instant case will be decided the same way as was the nearest case. SHYSTER then uses information about these nearest cases to construct a report. The report that SHYSTER generates makes a prediction and justifies that prediction by reference only to cases and their similarities and differences: the calculations that SHYSTER performs in coming to its opinion do not appear in that opinion. Safeguards are employed to warn users if SHYSTER doubts the veracity of its advice. == Results == SHYSTER was tested in four different and disparate areas of case law. Four specifications were written, each representing an area of Australian law: an aspect of the law of trover; the meaning of "authorization [sic]" in copyright law of Australia; the categorisation of employment contracts; and the implication of natural justice in administrative decision-making. SHYSTER was evaluated under five headings: its usefulness, its generality, the quality of its advice, its limitations, and possible enhancements that could be made to it. Despite its simple knowledge representation structure, it has shown itself capable of producing good advice, and its simple structure has facilitated the specification of different areas of law. Appreciating the difficulties encountered by legal expert systems developers in adequately representing legal knowledge can assist in appreciating the shortcomings of digital rights management technologies. Some academics believe future digital rights management systems may become sophisticated enough to permit exceptions to copyright law. To this end SHYSTER's attempt to model "authorization [sic]" in the Copyright Act can be viewed as pioneering work in this field. The term "authorization [sic]" is undefined in the Copyright Act. Consequently, a number of cases have been before the courts seeking answers as to what conduct amounts to authorisation. The main contexts in which the issue has arisen are analogous to permitted exceptions to copyright currently prevented by most digital rights management technologies: "home taping of recorded materials, photocopying in educational institutions and performing works in public". When applied to one case concerning compact cassettes, SHYSTER successfully agreed that Amstrad did not authorise the infringement. 'shyster-myci'n Popple highlighted the most obvious avenue of future research using SHYSTER as the development of a rule-based system, and the linking together of that rule-based system with the existing case-based system to form a hybrid system. This intention was eventually realised by Thomas O’Callaghan, the creator of SHYSTER-MYCIN: a hybrid legal expert system first presented at ICAIL '03, 24–28 June 2003 in Edinburgh, Scotland. MYCIN is an existing medical expert system, which was adapted for use with SHYSTER. MYCIN’s controversial "certainty factor" is not used in SHYSTER-MYCIN. The reason for this is the difficulty in scientifically establishing how certain a fact is in a legal domain. The rule-based approach of the MYCIN part is used to reason with the provisions of an Act of Parliament only. This hybrid system enables the case-based system (SHYSTER) to determine open textured concepts when required by the rule-based system (MYCIN). The ultimate conclusion of this joint endeavour is that a hybrid approach is preferred in the creation of legal expert systems where "it is appropriate to use rule-based reasoning when dealing with statutes, and…case-based reasoning when dealing with cases".

    Read more →
  • India AI Impact Summit 2026

    India AI Impact Summit 2026

    The India AI Impact Summit 2026 (also abbreviated as the AI Impact Summit) was an international summit on artificial intelligence held at Bharat Mandapam, New Delhi, India, from 16 to 21 February 2026. It is the fourth in a series of global AI summits following the Bletchley Park AI Safety Summit in 2023, the AI Seoul Summit in 2024, and the AI Action Summit in Paris in 2025. Organised under the IndiaAI Mission by the Ministry of Electronics and Information Technology, it is the first summit in the series to be hosted by a Global South nation. This series of AI summits will continue with the AI Summit in Geneva to be hosted by Switzerland in 2027. The summit was inaugurated by Prime Minister Narendra Modi on 19 February 2026. The opening ceremony was also addressed by French President Emmanuel Macron and United Nations Secretary-General António Guterres. The summit was attended by over 20 heads of state and a delegation of global technology leaders including Sundar Pichai (Google), Sam Altman (OpenAI), and Demis Hassabis (DeepMind). The event faced criticism for organisational issues, misrepresentation of non-Indian products as Indian, and a perceived focus on trade fair activities over substantive governance. == Background == The AI Impact Summit was an international summit on artificial intelligence (AI) held in New Delhi from 16 to 20 February 2026. It followed the AI Action Summit in Paris in February 2025, the AI Seoul Summit in 2024 and the Bletchley Park AI Safety Summit in 2023. According to Crowell & Moring, the changing summit titles seemed to reflect a broader shift in focus away from AI safety and governance toward practical impact, implementation, and measurable outcomes. Ahead of the summit, an international panel of experts published the second International AI Safety Report. The summit was structured around three foundational pillars, termed "Sutras": People, Planet, and Progress. Seven thematic working groups were established to deliver outcomes across these pillars, covering AI for economic growth and social good; democratising AI resources; inclusion for social empowerment; safe and trusted AI; human capital; science; and resilience, innovation, and efficiency. == Programme == The summit ran over five days, later extended to six following overwhelming public response. Originally scheduled to conclude on 20 February, the event was extended to 21 February with expanded evening hours for the exhibition. === India AI Impact Expo === The India AI Impact Expo, inaugurated by Prime Minister Modi on 16 February, featured over 300 exhibitors from 30 countries across more than 10 thematic pavilions. Pavilions were organised across thematic zones aligned with the summit's three pillars, showcasing AI applications in healthcare, agriculture, education, and sustainable industry. === Leaders' Plenary and CEO Roundtable === The Leaders' Plenary on 19 February brought together heads of state, ministers, and representatives from multilateral institutions to outline national and global priorities on AI governance, infrastructure, and international cooperation. A CEO Roundtable, held the same evening, convened senior executives from global technology and industry firms with government leaders to discuss investment, research collaboration, and deployment of AI systems. === Research Symposium === A Research Symposium on AI and its Impact was held on 18 February, with the IIIT Hyderabad as knowledge partner. Discussions covered sovereign AI infrastructure, global adoption challenges, research breakthroughs, and policy priorities. == Participants == The summit drew delegations from over 100 countries, including more than 20 heads of state and 60 ministers. Notable attendees from the technology industry included Sundar Pichai (Google), Sam Altman (OpenAI), Dario Amodei (Anthropic), Demis Hassabis (Google DeepMind), and Mukesh Ambani (Reliance Industries). Representatives from multilateral institutions included Sangbu Kim of the World Bank. == Announcements and outcomes == === Indian AI models === Several Indian AI models and products were unveiled during the summit. Sarvam AI, an Indian AI laboratory, launched a new generation of large language models, including 30-billion and 105-billion parameter models using a mixture of experts architecture, as well as text-to-speech, speech-to-text, and vision models. Sarvam also introduced the Kaze smartglasses, described as the company's first hardware product, which Prime Minister Modi tested at the expo. The government-backed BharatGen Param2 model, a 17-billion parameter model supporting 22 Indian languages with multimodal capabilities, was also launched at the summit. === Infrastructure commitments === Union Minister Ashwini Vaishnaw outlined India's "whole-of-nation" AI strategy, describing plans to build a "frugal, sovereign and scalable" AI ecosystem. The government announced plans to add more than 20,000 GPUs to India's existing base of 38,000 under the IndiaAI Compute Portal. Microsoft announced at the summit that it was on track to invest US$50 billion by the end of the decade to bring AI to lower-income countries. Goa reaffirmed its commitment to artificial intelligence at the India AI Impact Summit 2026. === Guinness World Record === During the summit, India set a Guinness World Record for the most pledges received for an AI responsibility campaign in 24 hours, with 250,946 valid pledges collected between 16 and 17 February 2026. The campaign, conducted in partnership with Intel India as part of the IndiaAI Mission, exceeded its initial target of 5,000 pledges. == Controversies and criticisms == === Galgotias University incident === On 18 February, Galgotias University faced widespread criticism after a representative presented a robot dog at the university's exhibition pavilion as an indigenous development. Social media users identified the robot as the Unitree Go2, a commercially available product manufactured by Chinese company Unitree Robotics. IT Secretary S. Krishnan stated that the government did not want exhibitors to showcase items that were not their own, and the university was directed to vacate its stall. Galgotias University issued an apology, stating that the representative had been "ill-informed" and was not authorised to speak to the press. The incident drew political reactions, with the Indian National Congress using it to criticise the government. The controversy was amplified after Union IT Minister Ashwini Vaishnaw had earlier shared a video clip of the robot on social media, which was subsequently deleted. === Organisational issues === On day 1 of the Summit, Dhananjay Yadav, a Bengaluru-based entrepreneur had alleged that his product was stolen in the Summit. He called it as a pain for the people in an X post. He further wrote, "Think about this: We paid for flights, accommodation, logistics and even the booth. Only to see our wearables disappear inside a high-security zone". Later, the stolen devices were recovered by The Delhi Police. Bloomberg reported that delegates were left stranded without food or water during a security lockdown ahead of the Prime Minister's visit on 19 February. The summit venue was closed to the public on 19 February for the Prime Minister's visit, leading to criticism from attendees who had registered for that day. === Protests by the Indian Youth Congress (IYC) === On 20 February, some members of the Indian Youth Congress (IYC) carried out protests inside the venue with slogans such as "PM is compromised" and the criticism of the recent trade deal between India and the US. 4 of these members were sent to police custody by the court on 22 February. While Bharatiya Janta Party condemned these protests, with its spokesperson Shehzad Poonawalla saying, "From being anti-BJP, you have gone to being anti-national? If you have a problem with the BJP, then protest at the BJP office, Jantar Mantar, or outside the PM's office. But the people of the country and their alliance partners condemn them for their attempt to defame India in front of the entire world at the AI Summit." Congress leader Harish Rawat defended the protests, saying "it's also a fact that AI might become a tool in the hands of a few individuals… It's the opposition's job to warn against that… It's not the first time such international events have been opposed. I know how the BJP protested during the Commonwealth Games… To say that such opposition has happened for the first time is not correct. The BJP has been doing this while in the opposition." These protestors were granted bail by the Delhi high court on 2 March. == Reception and analysis == Bloomberg News reported that Prime Minister Modi used the summit to assert India's global AI ambitions following a challenging year in foreign policy. TechPolicy.Press published several critical analyses of the summit. One article argued that the summit's structure granted "multinational corporations parity with sovereign governments

    Read more →
  • Neural architecture search

    Neural architecture search

    Neural architecture search (NAS) is a technique for automating the design of artificial neural networks (ANN), a widely used model in the field of machine learning. NAS has been used to design networks that are on par with or outperform hand-designed architectures. Methods for NAS can be categorized according to the search space, search strategy and performance estimation strategy used: The search space defines the type(s) of ANN that can be designed and optimized. The search strategy defines the approach used to explore the search space. The performance estimation strategy evaluates the performance of a possible ANN from its design (without constructing and training it). NAS is closely related to hyperparameter optimization and meta-learning and is a subfield of automated machine learning (AutoML). == Reinforcement learning == Reinforcement learning (RL) can underpin a NAS search strategy. Barret Zoph and Quoc Viet Le applied NAS with RL targeting the CIFAR-10 dataset and achieved a network architecture that rivals the best manually-designed architecture for accuracy, with an error rate of 3.65, 0.09 percent better and 1.05x faster than a related hand-designed model. On the Penn Treebank dataset, that model composed a recurrent cell that outperforms LSTM, reaching a test set perplexity of 62.4, or 3.6 perplexity better than the prior leading system. On the PTB character language modeling task it achieved bits per character of 1.214. Learning a model architecture directly on a large dataset can be a lengthy process. NASNet addressed this issue by transferring a building block designed for a small dataset to a larger dataset. The design was constrained to use two types of convolutional cells to return feature maps that serve two main functions when convoluting an input feature map: normal cells that return maps of the same extent (height and width) and reduction cells in which the returned feature map height and width is reduced by a factor of two. For the reduction cell, the initial operation applied to the cell's inputs uses a stride of two (to reduce the height and width). The learned aspect of the design included elements such as which lower layer(s) each higher layer took as input, the transformations applied at that layer and to merge multiple outputs at each layer. In the studied example, the best convolutional layer (or "cell") was designed for the CIFAR-10 dataset and then applied to the ImageNet dataset by stacking copies of this cell, each with its own parameters. The approach yielded accuracy of 82.7% top-1 and 96.2% top-5. This exceeded the best human-invented architectures at a cost of 9 billion fewer FLOPS—a reduction of 28%. The system continued to exceed the manually-designed alternative at varying computation levels. The image features learned from image classification can be transferred to other computer vision problems. E.g., for object detection, the learned cells integrated with the Faster-RCNN framework improved performance by 4.0% on the COCO dataset. In the so-called Efficient Neural Architecture Search (ENAS), a controller discovers architectures by learning to search for an optimal subgraph within a large graph. The controller is trained with policy gradient to select a subgraph that maximizes the validation set's expected reward. The model corresponding to the subgraph is trained to minimize a canonical cross entropy loss. Multiple child models share parameters, ENAS requires fewer GPU-hours than other approaches and 1000-fold less than "standard" NAS. On CIFAR-10, the ENAS design achieved a test error of 2.89%, comparable to NASNet. On Penn Treebank, the ENAS design reached test perplexity of 55.8. == Evolution == An alternative approach to NAS is based on evolutionary algorithms, which has been employed by several groups. An Evolutionary Algorithm for Neural Architecture Search generally performs the following procedure. First a pool consisting of different candidate architectures along with their validation scores (fitness) is initialised. At each step the architectures in the candidate pool are mutated (e.g.: 3x3 convolution instead of a 5x5 convolution). Next the new architectures are trained from scratch for a few epochs and their validation scores are obtained. This is followed by replacing the lowest scoring architectures in the candidate pool with the better, newer architectures. This procedure is repeated multiple times and thus the candidate pool is refined over time. Mutations in the context of evolving ANNs are operations such as adding or removing a layer, which include changing the type of a layer (e.g., from convolution to pooling), changing the hyperparameters of a layer, or changing the training hyperparameters. On CIFAR-10 and ImageNet, evolution and RL performed comparably, while both slightly outperformed random search. == Bayesian optimization == Bayesian Optimization (BO), which has proven to be an efficient method for hyperparameter optimization, can also be applied to NAS. In this context, the objective function maps an architecture to its validation error after being trained for a number of epochs. At each iteration, BO uses a surrogate to model this objective function based on previously obtained architectures and their validation errors. One then chooses the next architecture to evaluate by maximizing an acquisition function, such as expected improvement, which provides a balance between exploration and exploitation. Acquisition function maximization and objective function evaluation are often computationally expensive for NAS, and make the application of BO challenging in this context. Recently, BANANAS has achieved promising results in this direction by introducing a high-performing instantiation of BO coupled to a neural predictor. == Hill-climbing == Another group used a hill climbing procedure that applies network morphisms, followed by short cosine-annealing optimization runs. The approach yielded competitive results, requiring resources on the same order of magnitude as training a single network. E.g., on CIFAR-10, the method designed and trained a network with an error rate below 5% in 12 hours on a single GPU. == Multi-objective search == While most approaches solely focus on finding architecture with maximal predictive performance, for most practical applications other objectives are relevant, such as memory consumption, model size or inference time (i.e., the time required to obtain a prediction). Because of that, researchers created a multi-objective search. LEMONADE is an evolutionary algorithm that adopted Lamarckism to efficiently optimize multiple objectives. In every generation, child networks are generated to improve the Pareto frontier with respect to the current population of ANNs. Neural Architect is claimed to be a resource-aware multi-objective RL-based NAS with network embedding and performance prediction. Network embedding encodes an existing network to a trainable embedding vector. Based on the embedding, a controller network generates transformations of the target network. A multi-objective reward function considers network accuracy, computational resource and training time. The reward is predicted by multiple performance simulation networks that are pre-trained or co-trained with the controller network. The controller network is trained via policy gradient. Following a modification, the resulting candidate network is evaluated by both an accuracy network and a training time network. The results are combined by a reward engine that passes its output back to the controller network. == One-shot models == RL or evolution-based NAS require thousands of GPU-days of searching/training to achieve state-of-the-art computer vision results as described in the NASNet, mNASNet and MobileNetV3 papers. To reduce computational cost, many recent NAS methods rely on the weight-sharing idea. In this approach, a single overparameterized supernetwork (also known as the one-shot model) is defined. A supernetwork is a very large Directed Acyclic Graph (DAG) whose subgraphs are different candidate neural networks. Thus, in a supernetwork, the weights are shared among a large number of different sub-architectures that have edges in common, each of which is considered as a path within the supernet. The essential idea is to train one supernetwork that spans many options for the final design rather than generating and training thousands of networks independently. In addition to the learned parameters, a set of architecture parameters are learnt to depict preference for one module over another. Such methods reduce the required computational resources to only a few GPU days. More recent works further combine this weight-sharing paradigm, with a continuous relaxation of the search space, which enables the use of gradient-based optimization methods. These approaches are generally referred to as differentiable NAS and have proven very efficient in exploring the search space of ne

    Read more →
  • Human visual system model

    Human visual system model

    A human visual system model (HVS model) is used by image processing, video processing and computer vision experts to deal with biological and psychological processes that are not yet fully understood. Such a model is used to simplify the behaviors of what is a very complex system. As our knowledge of the true visual system improves, the model is updated. Psychovisual study is the study of the psychology of vision. The human visual system model can produce desired effects in perception and vision. Examples of using an HVS model include color television, lossy compression, and Cathode-ray tube (CRT) television. Originally, it was thought that color television required too high a bandwidth for the then available technology. Then it was noticed that the color resolution of the HVS was much lower than the brightness resolution; this allowed color to be squeezed into the signal by chroma subsampling. Another example is lossy image compression, like JPEG. Our HVS model says we cannot see high frequency detail, so in JPEG we can quantize these components without a perceptible loss of quality. Similar concepts are applied in audio compression, where sound frequencies inaudible to humans are band-stop filtered. Several HVS features are derived from evolution when we needed to defend ourselves or hunt for food. We often see demonstrations of HVS features when we are looking at optical illusions. == Block diagram of HVS == == Assumptions about the HVS == Low-pass filter characteristic (limited number of rods in human eye): see Mach bands Lack of color resolution (fewer cones in human eye than rods) Motion sensitivity More sensitive in peripheral vision Stronger than texture sensitivity, e.g. viewing a camouflaged animal Texture stronger than disparity – 3D depth resolution does not need to be so accurate Integral Face recognition (babies smile at faces) Depth inverted face looks normal (facial features overrule depth information) Upside down face with inverted mouth and eyes looks normal == Examples of taking advantage of an HVS model == Flicker frequency of film and television using persistence of vision to fool viewer into seeing a continuous image Interlaced television painting half images to give the impression of a higher flicker frequency Color television (chrominance at half resolution of luminance corresponding to proportions of rods and cones in eye) Image compression (difficult to see higher frequencies more harshly quantized) Motion estimation (use luminance and ignore color) Watermarking and Steganography

    Read more →
  • Really Simple Licensing

    Really Simple Licensing

    Really Simple Licensing (RSL) is an open content licensing standard that allows web publishers to set terms for web crawlers gathering training data for generative AI use. It was launched on September 10, 2025 and is managed by the nonprofit RSL Collective, co-founded by RSS co-creator Eckart Walther and former Ask.com CEO Doug Leeds. Participating companies at launch include Reddit, Yahoo, and Medium. Publishers can implement the RSL standard by adding licensing terms to their robots.txt files.

    Read more →
  • Tales from the Loop (role-playing game)

    Tales from the Loop (role-playing game)

    Tales from the Loop (Swedish: Ur Varselklotet), subtitled "Roleplaying in the '80s That Never Was", is an alternative history science fiction tabletop role-playing game published in 2017 by Free League Publishing, the international arm of Swedish game and book publisher Fria Ligan AB, and Modiphius Entertainment. The game, based on the art of Simon Stålenhag, envisions an alternative world where a group of bored and ignored preteens and teens solve mysteries caused by new technology near their hometown. == Description == === Setting === Tales from the Loop is set in an alternative history world taken from the artwork of Simon Stålenhag. According to this alternative timeline, back in the 1940s, research began on particle accelerators. In the 1960s, two massive underground particle accelerators were built in Sweden and Colorado with the promise of a harvest of technological marvels that would change everyone's lives. Tales from the Loop is set twenty years later, in the late 1980s, and the better life has not materialized. Although the particle accelerators have created robots and large skyships, the detritus of failed experiments and the ruins of abandoned high tech company buildings litter the landscape. Generally the life of the average family has not changed for the better. A campaign can either be set in the Mälaren Islands, west of the Swedish capital of Stockholm, or in a city in the Southwest United States that resembles Boulder City, Nevada. There is also a step-by-step guide for the gamemaster to use their own hometown. === Character generation === Player characters are preteens and young teenagers age 10–15 who live in a society where they are bored and largely left to themselves. Players can choose archetypes for their characters including Bookworm, Jock, Troublemaker, Popular Kid and Weirdo. Unlike most role-playing games, characters in Tales from the Loop cannot be killed, although in an ongoing campaign or due to an in-game effect, they are removed from the game if they reach the age of sixteen. === Game system === The game uses the Year Zero Engine first developed by Tomas Härenstam for the post-apocalyptic role-playing game Mutant: Year Zero. (Härenstam served as the editor and project manager for Tales from the Loop.) Problems are resolved by rolling a pool of six-sided dice, with any 6 rolled marking success. Attributes and skills (Sneak, Force, Move, Build, Tinker, Calculate, Contact, Charm, Lead, Investigate, Comprehend, and Empathize) may allow the player to add more dice to the dice pool, increasing the chances of success. However, if a character has earned a condition such as Scared or Injured, dice are removed from the dice pool. === Gameplay === The game principles are that life for the characters is dull and boring, but the area around the town is full of wonderful, mysterious things. An adventure is set up as a Mystery, and in order to successfully resolve the Mystery, characters must overcome a series of Troubles, which can range from having to be home by a certain time to dealing with a bully to disarming or otherwise overcoming a booby-trap on a door that must be opened. Each Mystery is played as a series of scenes, much like a TV drama. Although the gamemaster leads the players into the Mystery, each scene is set collaboratively with the players before action continues. As critic Jukka Kauppinen noted, "The players and the gamemaster take turns verbally staging a new scene — where we are, what it's like there — and only then what we do." === Campaign === The book presents a chronologically-linked set of four Mysteries called "The Four Seasons of Mad Science" that take place over a calendar year: "Summer Break and Killer Birds": The Kids hears pigeons having a conversation and investigate "Grown-Up Attraction": Adults start disappearing without any sign of struggle. "Creatures from the Cretaceous": The search for a missing dog leads to the discovery of creatures that don't belong in our time "I, Wagner": The Kids discover a body in a stream, and are drawn into a Mystery with robots and humans that may affect them closely. == Publication history == In 2017, Swedish artist Simon Stålenhag was raising money on Kickstarter to publish a book of his art titled Tales from the Loop. One of the stretch goals offered was the creation of a role-playing game. A second Kickstarter campaign to publish the role-playing game was initiated by Fria Ligan AB, who surpassed their crowdfunding goal and raised a total of 3,745,896 kr from 5,600 backers. The role-playing game Tales from the Loop was subsequently published as a 184-page hardcover book in 2017 by Free League Publishing, the international arm of Swedish game and book publisher Fria Ligan AB, and Modiphius Entertainment. Cover art and interior art were by Stålenhag, and cartography was by Christian Granath. A stand-alone expansion, Things from the Flood (Swedish: Flodskörden), based on Stålenhag's art book of the same name, was created by Nils Hintze, Rickard Antroia, and Tomas Härenstam. The 216-page hardcover book was published in 2019 with cover art by Stålenhag, interior art by Stålenhag and Reine Rosenberg, and cartography by Christian Granath. In 2020, the setting of the role-playing game was transferred to the TV series Tales from the Loop developed by Nathanial Halpern and Simon Stålenhag. The series tells eight stories of children's encounters with strange technology. == Reception == Shut Up & Sit Down praised Tales from the Loop for its comfortable, contemporary setting, simple rules that make the game easy to run, and the alternation between sci-fi and the kids' lives, but criticized the Type system for characters, noting "a suggested 'Pride' for the Weirdo involved being homosexual –– the only mention of queerness in the entire game. Those of us who identify as GLBTQ bristled at that: why was only the Weirdo queer, with queerness as a (possibly secret) Pride? Why not more fully address being a GLBTQ kid in the 1980s?" The review concluded, "For new RPG players, Tales is a decent game that you'll enjoy and that will make your heart burst. But you need an experienced GM who’s able to either alter the book’s mysteries or create their own, and who can put in work when poor dice rolls hold the players back." Rob Weiland of Geek & Sundry named Tales from the Loop 2017's best RPG release and praised Stålenhag's art, the collaborative nature between the GM and players, and the simplicity of running the game. Weiland concluded, "It has a simple system that is easy to explain but holds up under several plays. It has a setting that’s immediately evocative but also leaves plenty of room for GMs to build out their own world. It offers players a chance to experience the rush of memory, the pain of childhood and the wonder of movies." In a review of Tales from the Loop in Black Gate, Andrew Zimmerman Jones said, "Though not based directly on an established franchise, it draws richly from elements of popular culture that will make it resonate with many players. The focus on narrative play also means it’s a good game for people who aren’t necessarily big into learning a ton of new rules." Jukka Kauppinen, writing for the Finnish games magazine Skrolli, called the game, "downright delicious in its diversity. The science fiction world created by the Swedish artist Simon Stälenhag is, after all, both delightful vintage and tickling novelty." Kauppinen concluded, "This mutual storytelling and interaction makes this game more of a campfire circle than a traditional role-playing game. At the same time, its setting in the real world, tinged with science fiction and even horror, creates a delicious and unique adventure environment." In his 2023 book Monsters, Aliens, and Holes in the Ground, RPG historian Stu Horvath noted that the game system "pushes the players to constantly reevaluate their characters' relationships with the everyday world, for better or worse. It won't be long before navigating entanglements with parents, teachers, siblings and bullies proves just as risky to the characters, and central to the players' experience, as trying to find out what happened with the time portal or dealing with a rampaging robot." Horvath concluded, "The appeal of Tales from the Loop is Stålenhag's deep shadows and purple dusks. They hide the dangers and mysteries that often act [as] an escape hatch, a way to avoid prosaic problems." == Awards == At the 2017 Golden Geek Awards, Tales of the Loop won "RPG of the Year", and was a finalist for " Best RPG Artwork/Presentation" At the 2017 ENnie Awards, Tales from the Loops won five Gold Medals: Product of the Year Best Writing Best Setting Best Game Best Art, Interior

    Read more →
  • Ensemble averaging (machine learning)

    Ensemble averaging (machine learning)

    In machine learning, ensemble averaging is the process of creating multiple models (typically artificial neural networks) and combining them to produce a desired output, as opposed to creating just one model. Ensembles of models often outperform individual models, as the various errors of the ensemble constituents "average out". == Overview == Ensemble averaging is one of the simplest types of committee machines. Along with boosting, it is one of the two major types of static committee machines. In contrast to standard neural network design, in which many networks are generated but only one is kept, ensemble averaging keeps the less satisfactory networks, but with less weight assigned to their outputs. The theory of ensemble averaging relies on two properties of artificial neural networks: In any network, the bias can be reduced at the cost of increased variance In a group of networks, the variance can be reduced at no cost to the bias. This is known as the bias–variance tradeoff. Ensemble averaging creates a group of networks, each with low bias and high variance, and combines them to form a new network which should theoretically exhibit low bias and low variance. Hence, this can be thought of as a resolution of the bias–variance tradeoff. The idea of combining experts can be traced back to Pierre-Simon Laplace. == Method == The theory mentioned above gives an obvious strategy: create a set of experts with low bias and high variance, and average them. Generally, what this means is to create a set of experts with varying parameters; frequently, these are the initial synaptic weights of a neural network, although other factors (such as learning rate, momentum, etc.) may also be varied. Some authors recommend against varying weight decay and early stopping. The steps are therefore: Generate N experts, each with their own initial parameters (these values are usually sampled randomly from a distribution) Train each expert separately Combine the experts and average their values. Alternatively, domain knowledge may be used to generate several classes of experts. An expert from each class is trained, and then combined. A more complex version of ensemble average views the final result not as a mere average of all the experts, but rather as a weighted sum. If each expert is y i {\displaystyle y_{i}} , then the overall result y ~ {\displaystyle {\tilde {y}}} can be defined as: y ~ ( x ; α ) = ∑ j = 1 p α j y j ( x ) {\displaystyle {\tilde {y}}(\mathbf {x} ;\mathbf {\alpha } )=\sum _{j=1}^{p}\alpha _{j}y_{j}(\mathbf {x} )} where α {\displaystyle \mathbf {\alpha } } is a set of weights. The optimization problem of finding alpha is readily solved through neural networks, hence a "meta-network" where each "neuron" is in fact an entire neural network can be trained, and the synaptic weights of the final network is the weight applied to each expert. This is known as a linear combination of experts. It can be seen that most forms of neural network are some subset of a linear combination: the standard neural net (where only one expert is used) is simply a linear combination with all α j = 0 {\displaystyle \alpha _{j}=0} and one α k = 1 {\displaystyle \alpha _{k}=1} . A raw average is where all α j {\displaystyle \alpha _{j}} are equal to some constant value, namely one over the total number of experts. A more recent ensemble averaging method is negative correlation learning, proposed by Y. Liu and X. Yao. This method has been widely used in evolutionary computing. == Benefits == The resulting committee is almost always less complex than a single network that would achieve the same level of performance The resulting committee can be trained more easily on smaller datasets The resulting committee often has improved performance over any single model The risk of overfitting is lessened, as there are fewer parameters (e.g. neural network weights) which need to be set.

    Read more →
  • Gold (linker)

    Gold (linker)

    In software engineering, gold is a linker for ELF files. It became an official GNU package and was added to binutils in March 2008 and first released in binutils version 2.19. gold was developed by Ian Lance Taylor and a small team at Google. The motivation for writing gold was to make a linker that is faster than the GNU linker, especially for large applications coded in C++. Unlike the GNU linker, gold does not use the BFD library to process object files. While this limits the object file formats it can process to ELF only, it is also claimed to result in a cleaner and faster implementation without an additional abstraction layer. The author cited complete removal of BFD as a reason to create a new linker from scratch rather than incrementally improve the GNU linker. This rewrite also fixes some bugs in old ld that break ELF files in various minor ways. To specify gold in a makefile, one sets the LD or LD environment variable to ld.gold. To specify gold through a compiler option, one can use the gcc option -fuse-ld=gold. Fedora has moved gold from binutils into its own package due to concerns it is suffering from bitrot after Google's interest has moved to LLVM. In particular, gold does not read LDFLAGS variable, so cannot see libraries in folders like /usr/local/lib. On 2025-02-02 the 2.44 version of GNU Binutils removed gold from the default source distribution and into a separate package, stating that "the gold linker is now deprecated and will eventually be removed unless volunteers step forward and offer to continue development and maintenance".

    Read more →
  • Content-based image retrieval

    Content-based image retrieval

    Content-based image retrieval, also known as query by image content (QBIC) and content-based visual information retrieval (CBVIR), is the application of computer vision techniques to the image retrieval problem, that is, the problem of searching for digital images in large databases (see this survey for a scientific overview of the CBIR field). Content-based image retrieval is opposed to traditional concept-based approaches (see Concept-based image indexing). "Content-based" means that the search analyzes the contents of the image rather than the metadata such as keywords, tags, or descriptions associated with the image. The term "content" in this context might refer to colors, shapes, textures, or any other information that can be derived from the image itself. CBIR is desirable because searches that rely purely on metadata are dependent on annotation quality and completeness. == Comparison with metadata searching == An image meta search requires humans to have manually annotated images by entering keywords or metadata in a large database, which can be time-consuming and may not capture the keywords desired to describe the image. The evaluation of the effectiveness of keyword image search is subjective and has not been well-defined. In the same regard, CBIR systems have similar challenges in defining success. "Keywords also limit the scope of queries to the set of predetermined criteria." and, "having been set up" are less reliable than using the content itself. == History == The term "content-based image retrieval" seems to have originated in 1992 when it was used by Japanese Electrotechnical Laboratory engineer Toshikazu Kato to describe experiments into automatic retrieval of images from a database, based on the colors and shapes present. Since then, the term has been used to describe the process of retrieving desired images from a large collection on the basis of syntactical image features. The techniques, tools, and algorithms that are used originate from fields such as statistics, pattern recognition, signal processing, and computer vision. === QBIC - Query By Image Content === The earliest commercial CBIR system was developed by IBM and was called QBIC (Query By Image Content). Recent network- and graph-based approaches have presented a simple and attractive alternative to existing methods. While the storing of multiple images as part of a single entity preceded the term BLOB (Binary Large OBject), the ability to fully search by content, rather than by description, had to await IBM's QBIC. === VisualRank === == Technical progress == The interest in CBIR has grown because of the limitations inherent in metadata-based systems, as well as the large range of possible uses for efficient image retrieval. Textual information about images can be easily searched using existing technology, but this requires humans to manually describe each image in the database. This can be impractical for very large databases or for images that are generated automatically, e.g. those from surveillance cameras. It is also possible to miss images that use different synonyms in their descriptions. Systems based on categorizing images in semantic classes like "cat" as a subclass of "animal" can avoid the miscategorization problem, but will require more effort by a user to find images that might be "cats", but are only classified as an "animal". Many standards have been developed to categorize images, but all still face scaling and miscategorization issues. Initial CBIR systems were developed to search databases based on image color, texture, and shape properties. After these systems were developed, the need for user-friendly interfaces became apparent. Therefore, efforts in the CBIR field started to include human-centered design that tried to meet the needs of the user performing the search. This typically means inclusion of: query methods that may allow descriptive semantics, queries that may involve user feedback, systems that may include machine learning, and systems that may understand user satisfaction levels. == Techniques == Many CBIR systems have been developed, but as of 2006, the problem of retrieving images on the basis of their pixel content remains largely unsolved. Different query techniques and implementations of CBIR make use of different types of user queries. === Query By Example === QBE (Query By Example) is a query technique that involves providing the CBIR system with an example image that it will then base its search upon. The underlying search algorithms may vary depending on the application, but result images should all share common elements with the provided example. Options for providing example images to the system include: A preexisting image may be supplied by the user or chosen from a random set. The user draws a rough approximation of the image they are looking for, for example with blobs of color or general shapes. This query technique removes the difficulties that can arise when trying to describe images with words. === Semantic retrieval === Semantic retrieval starts with a user making a request like "find pictures of Abraham Lincoln". This type of open-ended task is very difficult for computers to perform - Lincoln may not always be facing the camera or in the same pose. Many CBIR systems therefore generally make use of lower-level features like texture, color, and shape. These features are either used in combination with interfaces that allow easier input of the criteria or with databases that have already been trained to match features (such as faces, fingerprints, or shape matching). However, in general, image retrieval requires human feedback in order to identify higher-level concepts. === Relevance feedback (human interaction) === Combining CBIR search techniques available with the wide range of potential users and their intent can be a difficult task. An aspect of making CBIR successful relies entirely on the ability to understand the user intent. CBIR systems can make use of relevance feedback, where the user progressively refines the search results by marking images in the results as "relevant", "not relevant", or "neutral" to the search query, then repeating the search with the new information. Examples of this type of interface have been developed. === Iterative/machine learning === Machine learning and application of iterative techniques are becoming more common in CBIR. === Other query methods === Other query methods include browsing for example images, navigating customized/hierarchical categories, querying by image region (rather than the entire image), querying by multiple example images, querying by visual sketch, querying by direct specification of image features, and multimodal queries (e.g. combining touch, voice, etc.) == Content comparison using image distance measures == The most common method for comparing two images in content-based image retrieval (typically an example image and an image from the database) is using an image distance measure. An image distance measure compares the similarity of two images in various dimensions such as color, texture, shape, and others. For example, a distance of 0 signifies an exact match with the query, with respect to the dimensions that were considered. As one may intuitively gather, a value greater than 0 indicates various degrees of similarities between the images. Search results then can be sorted based on their distance to the queried image. Many measures of image distance (Similarity Models) have been developed. === Color === Computing distance measures based on color similarity is achieved by computing a color histogram for each image that identifies the proportion of pixels within an image holding specific values. Examining images based on the colors they contain is one of the most widely used techniques because it can be completed without regard to image size or orientation. However, research has also attempted to segment color proportion by region and by spatial relationship among several color regions. === Texture === Texture measures look for visual patterns in images and how they are spatially defined. Textures are represented by texels which are then placed into a number of sets, depending on how many textures are detected in the image. These sets not only define the texture, but also where in the image the texture is located. Texture is a difficult concept to represent. The identification of specific textures in an image is achieved primarily by modeling texture as a two-dimensional gray level variation. The relative brightness of pairs of pixels is computed such that degree of contrast, regularity, coarseness and directionality may be estimated. The problem is in identifying patterns of co-pixel variation and associating them with particular classes of textures such as silky, or rough. Other methods of classifying textures include: Co-occurrence matrix Laws texture energy Wavelet transform Orthogonal transforms (discrete Chebyshev moments) =

    Read more →
  • Tilly Norwood

    Tilly Norwood

    Tilly Norwood is a character created using generative artificial intelligence in 2025 by Xicoia, the AI division of Particle6 Group, a production company founded by Eline Van der Velden. "AI Commissioner", the first project to feature the Norwood character, was criticised by reviewers for The Guardian, PC Gamer, and The A.V. Club. A press release that talent agencies expressed interest in representing the character attracted strong criticism from Hollywood actors and firms, prompting allegations of personality rights violations and arguments over the impact of the character on production costs in the media industry. == History == Norwood was created by Xicoia, which was founded in February 2025 as the artificial intelligence (AI) division of Particle6, a production company founded by Dutch actress and producer Eline Van der Velden in 2015. Van der Velden had previously starred in a satirical comedy series for BBC Three based around her character Miss Holland, whom she created in 2012 as a parody of beauty standards. She stated that the process of creating Norwood took "a long time" and compared the process to that of writers creating characters. An Instagram account under Norwood's name, with posts dating back to 6 May 2025, had gained 50,000 followers by October 3, and featured AI-generated modelling shots, selfies, and epic film scenes. Van der Velden stated in July 2025 that she intended Norwood to be the next Scarlett Johansson or Natalie Portman and later said that audiences were more interested in a film's story than whether its actors were real. Particle6 has claimed that using Norwood could cut production costs by 90%. On 30 July 2025, a comedy sketch named "AI Commissioner" was released, featuring Norwood as an "actress" along with other AI-generated characters. It was created with ten AI software tools, with a script generated by ChatGPT. Stuart Heritage of The Guardian described it as technically competent but "relentlessly unfunny to watch", with "sloppily written, woodenly delivered dialogue", and that Norwood's teeth kept "blurring into a single white block." Joshua Wolens of PC Gamer wrote that Norwood's exaggerated mouth movements gave the impression "that her skeleton was about to leave her body", while William Hughes of The A.V. Club wrote that the sketch's attempt at mimicking human body and mouth movements produced "such a hideous uncanny valley effect" that it gave them "a full-on case of the screaming fantods". By October 2, the sketch had been viewed more than 700,000 times on YouTube. Xicoia was officially announced on 27 September 2025, at the Zurich Summit, part of the Zurich Film Festival; there, van der Velden unveiled Norwood and later joined a panel with Verena Puhm, head of Luma AI's Studio Dream Lab LA. They suggested that media companies were quietly embracing AI and that public announcements of AI-generated works were imminent. Van der Velden claimed that studios had dropped their objections by May after being opposed in February, and that multiple talent agencies were considering representing Norwood. The latter claim drew heightened attention to the character and was printed as fact by Deadline under the headline "Talent Agents Circle AI Actress Tilly Norwood." The report caused controversy, with Vulture describing the reaction to it as "Hollywood [lurching] into a fresh wave of existential panic" while being critical of Deadline's reporting, writing that "when Deadline called it a 'revelation' and published the supposed interest as fact without verification, [it] metastasized into a full-fledged cyberpunk news cycle", and that "by Tuesday, it had grown like wildfire." By September 2025, AI-generated videos had been released depicting Norwood on a red carpet, crying on the sofa of The Graham Norton Show, and starring in mock trailers for sci-fi, fantasy, horror, and action films. Later that month, actresses Melissa Barrera, Kiersey Clemons, and Natasha Lyonne suggested boycotting any agency who signed Norwood, while Mara Wilson asked why none of the "hundreds of living young women whose faces were composited together" to create Norwood could be hired instead. Also around this time, Emily Blunt described Norwood as "really, really scary", and Sophie Turner, Toni Collette, Ralph Ineson, and Ariel Winter also expressed disapproval, while Lukas Gage, Odessa A'zion, and Trace Lysette joked about having supposedly worked with Norwood and finding her incompetent and unpleasant to work with, with Gage claiming that "She was a nightmare to work with!" and "She couldn't hit her mark and she was late!" and Lysette adding "She cut me in line at lunch one day and didn't even say excuse me. She won't get far." Jenelle Riley, Nicholas Alexander Chavez, and the American union SAG-AFTRA stated that they do not consider Norwood an actress. The Gersh Agency and WME both announced that they would not sign Norwood. Whoopi Goldberg and Charlie Fink expressed scepticism that AI could replace jobs. Esquire UK reported that a post on Deadline's Instagram account about Norwood also sparked "varying levels of disgust and outrage" in its comments section from Adelaide Kane, Eiza González, Katie Cassidy, Jewel Staite, Lucy Hale, Stephen Sean Ford, and others, singling out González's comment, saying "Shame on whoever is trying to normalize this. Horrific and terrifying." Actor Bronson Pinchot expressed concern that Norwood could take his job. The British union Equity and the Canadian union ACTRA also condemned Norwood. Following this criticism, Van der Velden released a statement claiming Norwood was "not a replacement for a human being, but a creative work." She also denied that a £120,000 grant from the British Film Institute to fund Particle6 had been used to create Norwood, stating that Norwood had been a self-funded project solely for Xicoia. In late October, businessman Kevin O'Leary, while advocating for the use of AI to replace background actors, stated that they could be replaced with "100 Norwell Tillies" without being able to tell the difference. Ryan Reynolds and a real woman named Natalie "Tilly" Norwood also starred in an advertisement for Mint Mobile's internet service provider Minternet that mocked the character of Norwood. In November 2025, Van der Velden stated in an interview with Deadline that she planned to create 40 further "very diverse" characters alongside Norwood in order to expand the character's "whole universe". Also that month, actress Jameela Jamil criticized the idea of Norwood as "deeply disturbing" for being "a teenage-looking girl who can't say no to a type of sex scene" or "advocate for herself". Van der Velden announced later that month that Particle6 would be producing the History Channel's Streets of the Past, a Dutch documentary series which would be hosted by reality television personality Corjan Mol and would use AI to recreate historical scenes. In March 2026, a music video titled "Take The Lead" featuring Norwood was released on YouTube. It addressed the backlash of Norwood's creation by opening with the lyrics: "When they talk about me, they don't see/ The human spark, the creativity," and, "I'm just a tool, but I've got life." It also featured a disclaimer that says: "made by 18 real humans — from production designers to costume designers to prompters, editors and an actor." The vocals were generated by Suno. == Commentary == Charles Pulliam-Moore of The Verge argued that Norwood's introduction was a stunt to normalize "AI actors" despite Norwood essentially being a digital puppet. Straight Arrow News compared Tilly Norwood to Aki Ross, a CGI character from 2001 that was similarly intended to become a "digital star" and appear in multiple films, while Nicholas Schrivens, writing for The Conversation, likened Norwood to the posthumous use of footage of Carrie Fisher as Princess Leia for Star Wars: The Rise of Skywalker in 2019 and the Los Angeles Times likened Norwood to Hatsune Miku. Scrivens also wrote that "no AI creation has achieved the media cut-through that Tilly has". Moises Mendez II of Out dismissed this as "vapid bullshit", writing, "Nobody wants AI actresses." Scottish actress Briony Monroe alleged that Norwood had been modeled after her likeness and mannerisms, and stated that she was consulting Equity regarding the matter. Musician Stella Hennen said in a viral TikTok video, which was uploaded in October 2025 and featured a side-by-side comparison between herself and Norwood, that Norwood was her "doppleganger". On April 14, 2026, Marie Claire published an article titled "Is Tilly Norwood the Most Dangerous 'Actress' in Hollywood?", though it noted that AI-generated characters are "still not very good at, well, acting," "audiences have not been kind to AI-led productions," and "Norwood's 'performances' have already faced negative reviews as well". The University of Southern California's Entertainment Technology Center's AI media director Yves Bergquist dismissed th

    Read more →
  • The Old Axolotl

    The Old Axolotl

    The Old Axolotl (Polish: Starość aksolotla) is a 2015 digital-only novel by Polish science-fiction author Jacek Dukaj. The novel was released in Polish on March 10, 2015, and shortly afterward, on March 24 that year, in English (translated by Stanley Bill). It has been described as "an experiment in reading (and creating) the electronic literature of the future". It is Dukaj's first novel to be published in English, though several of his short stories (The Golden Galley, 1996, The Iron General, 2010, The Apocrypha of Lem, 2011) have been translated prior to this. The novel has inspired two Netflix original series: the 2020 Belgian Into the Night, and its 2022 Turkish language spin-off Yakamoz S-245. == Plot == The novel presents a post-apocalyptic, cyberpunk vision of Earth where biological life has been wiped out, inhabited by robots and mechs, many of which are humans whose consciousness has been digitized in the wake of an extinction event. == Significance and analysis == The novel is an example of electronic literature, available only in digital formats, and has no traditional paper version. It was designed from the beginning not only to incorporate more traditional elements such as illustrations, but also hypertext, and 3D-printable models of main robotic characters designed by Alex Jaeger, the art director of Transformers films. The novel composition is layered, with the narrative layer, an encyclopedic/hyperlinked footnote layer, and a multimedia layer, including illustrations and a short promotional video by the Oscar-nominated Platige Image studio. One of the novel's central questions is: "What does it mean to be human?" Other subjects include post humanism and other "staples of cyberpunk and related genres, such as the artificial intelligence". The novel is representative of Dukaj's prose, posing philosophical questions about the future of man and technology. The author explained that: "stories such as The Old Axolotl that model an ‘escape from the body’ are born out of a sense of progress as a process of ‘de-animalising’ human beings through science. This has its origin in the pre-Enlightenment intuition of ‘liberation from nature’. For one of the last shackles of nature is corporeality itself, the limitations of our physicality." The other major element of the novel is Dukaj's attempts to introduce the reader to the new style of electronic literature. The novel was nominated for the 2016 Janusz A. Zajdel Award.

    Read more →
  • Image tracing

    Image tracing

    In computer graphics, image tracing, raster-to-vector conversion or raster vectorization is the conversion of raster graphics into vector graphics. == Background == An image does not have any structure: it is just a collection of marks on paper, grains in film, or pixels in a bitmap. While such an image is useful, it has some limits. If the image is magnified enough, its artifacts appear. The halftone dots, film grains, and pixels become apparent. Images of sharp edges become fuzzy or jagged. See, for example, pixelation. Ideally, a vector image does not have the same problem. Edges and filled areas are represented as mathematical curves or gradients, and they can be magnified arbitrarily (though of course the final image must also be rasterized in to be rendered, and its quality depends on the quality of the rasterization algorithm for the given inputs). The task in vectorization is to convert a two-dimensional image into a two-dimensional vector representation of the image. It is not examining the image and attempting to recognize or extract a three-dimensional model that may be depicted; i.e. it is not a vision system. For most applications, vectorization also does not involve optical character recognition; characters are treated as lines, curves, or filled objects without attaching any significance to them. In vectorization, the shape of the character is preserved, so artistic embellishments remain. Vectorization is the inverse operation corresponding to rasterization, as integration is to differentiation. And, just as with these other operations, while rasterization is fairly straightforward and algorithmic, vectorization involves the reconstruction of lost information and therefore requires heuristic methods. Synthetic images such as maps, cartoons, logos, clip art, and technical drawings are suitable for vectorization. Those images could have been originally made as vector images because they are based on geometric shapes or drawn with simple curves. Continuous tone photographs (such as live portraits) are not good candidates for vectorization. The input to vectorization is an image, but an image may come in many forms such as a photograph, a drawing on paper, or one of several raster file formats. Programs that do raster-to-vector conversion may accept bitmap formats such as TIFF, BMP and PNG. The output is a vector file format. Common vector formats are SVG, DXF, EPS, EMF and AI. Vectorization can be used to update images or recover work. Personal computers often come with a simple paint program that produces a bitmap output file. These programs allow users to make simple illustrations by adding text, drawing outlines, and filling outlines with a specific color. Only the results of these operations (the pixels) are saved in the resulting bitmap; the drawing and filling operations are discarded. Vectorization can be used to recapture some of the information that was lost. Vectorization is also used to recover information that was originally in a vector format but has been lost or has become unavailable. A company may have commissioned a logo from a graphic arts firm. Although the graphics firm used a vector format, the client company may not have received a copy of that format. The company may then acquire a vector format by scanning and vectorizing a paper copy of the logo. == Process == Vectorization starts with an image. === Manual === The image can be vectorized manually. A person could look at the image, make some measurements, and then write the output file by hand. That was the case for the vectorization of a technical illustration about neutrinos. The illustration has a few geometric shapes and a lot of text; it was relatively easy to convert the shapes, and the SVG vector format allows the text (even subscripts and superscripts) to be entered easily. The original image did not have any curves (except for the text), so the conversion is straightforward. Curves make the conversion more complicated. Manual vectorization of complicated shapes can be facilitated by the tracing function built into some vector graphics editing programs. If the image is not yet in machine readable form, then it has to be scanned into a usable file format. Once there is a machine-readable bitmap, the image can be imported into a graphics editing program (such as Adobe Illustrator, CorelDRAW, or Inkscape). Then a person can manually trace the elements of the image using the program's editing features. Curves in the original image can be approximated with lines, arcs, and Bézier curves. An illustration program allows spline knots to be adjusted for a close fit. Manual vectorization is possible, but it can be tedious. Although graphics drawing programs have been around for a long time, artists may find the freehand drawing facilities awkward even when a drawing tablet is used. Instead of using a program, Pepper recommends making an initial sketch on paper. Instead of scanning the sketch and tracing it freehand in the computer, Pepper states: "Those proficient with a graphic tablet and stylus could make the following changes directly in CorelDRAW by using a scan of the sketch as an underlay and drawing over it. I prefer to use pen and ink, and a light table"; most of the final image was traced by hand in ink. Later the line-drawing image was scanned at 600 dpi, cleaned up in a paint program, and then automatically traced with a program. Once the black and white image was in the graphics program, some other elements were added and the figure was colored. Similarly, Ploch recreated a design from a digital photograph. The JPEG was imported and some "basic shapes" were traced by hand and colored in the graphics drawing program; more complex shapes were handled differently. Ploch used a bitmap editor to remove the background and crop the more complex image components. He then printed the image and traced it by hand onto tracing paper to get a clean black and white line drawing. That drawing was scanned and then vectorized with a program. === Automatic === Some programs automate the vectorization process. Example programs are Adobe Illustrator, Inkscape, Corel's PowerTRACE, and Potrace. Some of these programs have a command line interface while others are interactive that allow the user to adjust the conversion settings and view the result. Adobe Streamline is not only an interactive program, but it also allows a user to manually edit the input bitmap and the output curves. Corel's PowerTRACE is accessed through CorelDRAW; CorelDRAW can be used to modify the input bitmap and edit the output curves. Adobe Illustrator has a facility to trace individual curves. Automated programs can have mixed results. A program (PowerTRACE) was used to convert a PNG map to SVG. The program did a good job on the map boundaries (the most tedious task in the tracing) and the settings dropped out all the text (small objects). The text was manually re-inserted. Other conversions may not go as well. The results depend on having high-quality scans, reasonable settings, and good algorithms. Scanned images often have a lot of noise, which can require additional work to clean up. == Options == There are many different image styles and possibilities, and no single vectorization method works well on all images. Consequently, vectorization programs have many options that influence the result. One issue is what the predominant shapes are. If the image is of a fill-in form, then it will probably have just vertical and horizontal lines of a constant width. The program's vectorization should take that into account. On the other hand, a CAD drawing may have lines at any angle, there may be curved lines, and there may be several line weights (thick for objects and thin for dimension lines). Instead of (or in addition to) curves, the image may contain outlines filled with the same color. Adobe Streamline allows users to select a combination of line recognition (horizontal and vertical lines), centerline recognition, or outline recognition. Streamline also allows small outline shapes to be thrown out; the notion is such small shapes are noise. The user may set the noise level between 0 and 1000; an outline that has fewer pixels than that setting is discarded. Another issue is the number of colors in the image. Even images that were created as black on white drawings may end up with many shades of gray. Some line-drawing routines employ anti-aliasing; a pixel completely covered by the line will be black, but a pixel that is only partially covered will be gray. If the original image is on paper and is scanned, there is a similar result: edge pixels will be gray. Sometimes images are compressed (e.g., JPEG images), and the compression will introduce gray levels. Many of the vectorization programs will group same-color pixels into lines, curves, or outlined shapes. If each possible color is grouped into its object, there can be an enormous number of objects. Instead, the user is asked to s

    Read more →
  • Neural architecture search

    Neural architecture search

    Neural architecture search (NAS) is a technique for automating the design of artificial neural networks (ANN), a widely used model in the field of machine learning. NAS has been used to design networks that are on par with or outperform hand-designed architectures. Methods for NAS can be categorized according to the search space, search strategy and performance estimation strategy used: The search space defines the type(s) of ANN that can be designed and optimized. The search strategy defines the approach used to explore the search space. The performance estimation strategy evaluates the performance of a possible ANN from its design (without constructing and training it). NAS is closely related to hyperparameter optimization and meta-learning and is a subfield of automated machine learning (AutoML). == Reinforcement learning == Reinforcement learning (RL) can underpin a NAS search strategy. Barret Zoph and Quoc Viet Le applied NAS with RL targeting the CIFAR-10 dataset and achieved a network architecture that rivals the best manually-designed architecture for accuracy, with an error rate of 3.65, 0.09 percent better and 1.05x faster than a related hand-designed model. On the Penn Treebank dataset, that model composed a recurrent cell that outperforms LSTM, reaching a test set perplexity of 62.4, or 3.6 perplexity better than the prior leading system. On the PTB character language modeling task it achieved bits per character of 1.214. Learning a model architecture directly on a large dataset can be a lengthy process. NASNet addressed this issue by transferring a building block designed for a small dataset to a larger dataset. The design was constrained to use two types of convolutional cells to return feature maps that serve two main functions when convoluting an input feature map: normal cells that return maps of the same extent (height and width) and reduction cells in which the returned feature map height and width is reduced by a factor of two. For the reduction cell, the initial operation applied to the cell's inputs uses a stride of two (to reduce the height and width). The learned aspect of the design included elements such as which lower layer(s) each higher layer took as input, the transformations applied at that layer and to merge multiple outputs at each layer. In the studied example, the best convolutional layer (or "cell") was designed for the CIFAR-10 dataset and then applied to the ImageNet dataset by stacking copies of this cell, each with its own parameters. The approach yielded accuracy of 82.7% top-1 and 96.2% top-5. This exceeded the best human-invented architectures at a cost of 9 billion fewer FLOPS—a reduction of 28%. The system continued to exceed the manually-designed alternative at varying computation levels. The image features learned from image classification can be transferred to other computer vision problems. E.g., for object detection, the learned cells integrated with the Faster-RCNN framework improved performance by 4.0% on the COCO dataset. In the so-called Efficient Neural Architecture Search (ENAS), a controller discovers architectures by learning to search for an optimal subgraph within a large graph. The controller is trained with policy gradient to select a subgraph that maximizes the validation set's expected reward. The model corresponding to the subgraph is trained to minimize a canonical cross entropy loss. Multiple child models share parameters, ENAS requires fewer GPU-hours than other approaches and 1000-fold less than "standard" NAS. On CIFAR-10, the ENAS design achieved a test error of 2.89%, comparable to NASNet. On Penn Treebank, the ENAS design reached test perplexity of 55.8. == Evolution == An alternative approach to NAS is based on evolutionary algorithms, which has been employed by several groups. An Evolutionary Algorithm for Neural Architecture Search generally performs the following procedure. First a pool consisting of different candidate architectures along with their validation scores (fitness) is initialised. At each step the architectures in the candidate pool are mutated (e.g.: 3x3 convolution instead of a 5x5 convolution). Next the new architectures are trained from scratch for a few epochs and their validation scores are obtained. This is followed by replacing the lowest scoring architectures in the candidate pool with the better, newer architectures. This procedure is repeated multiple times and thus the candidate pool is refined over time. Mutations in the context of evolving ANNs are operations such as adding or removing a layer, which include changing the type of a layer (e.g., from convolution to pooling), changing the hyperparameters of a layer, or changing the training hyperparameters. On CIFAR-10 and ImageNet, evolution and RL performed comparably, while both slightly outperformed random search. == Bayesian optimization == Bayesian Optimization (BO), which has proven to be an efficient method for hyperparameter optimization, can also be applied to NAS. In this context, the objective function maps an architecture to its validation error after being trained for a number of epochs. At each iteration, BO uses a surrogate to model this objective function based on previously obtained architectures and their validation errors. One then chooses the next architecture to evaluate by maximizing an acquisition function, such as expected improvement, which provides a balance between exploration and exploitation. Acquisition function maximization and objective function evaluation are often computationally expensive for NAS, and make the application of BO challenging in this context. Recently, BANANAS has achieved promising results in this direction by introducing a high-performing instantiation of BO coupled to a neural predictor. == Hill-climbing == Another group used a hill climbing procedure that applies network morphisms, followed by short cosine-annealing optimization runs. The approach yielded competitive results, requiring resources on the same order of magnitude as training a single network. E.g., on CIFAR-10, the method designed and trained a network with an error rate below 5% in 12 hours on a single GPU. == Multi-objective search == While most approaches solely focus on finding architecture with maximal predictive performance, for most practical applications other objectives are relevant, such as memory consumption, model size or inference time (i.e., the time required to obtain a prediction). Because of that, researchers created a multi-objective search. LEMONADE is an evolutionary algorithm that adopted Lamarckism to efficiently optimize multiple objectives. In every generation, child networks are generated to improve the Pareto frontier with respect to the current population of ANNs. Neural Architect is claimed to be a resource-aware multi-objective RL-based NAS with network embedding and performance prediction. Network embedding encodes an existing network to a trainable embedding vector. Based on the embedding, a controller network generates transformations of the target network. A multi-objective reward function considers network accuracy, computational resource and training time. The reward is predicted by multiple performance simulation networks that are pre-trained or co-trained with the controller network. The controller network is trained via policy gradient. Following a modification, the resulting candidate network is evaluated by both an accuracy network and a training time network. The results are combined by a reward engine that passes its output back to the controller network. == One-shot models == RL or evolution-based NAS require thousands of GPU-days of searching/training to achieve state-of-the-art computer vision results as described in the NASNet, mNASNet and MobileNetV3 papers. To reduce computational cost, many recent NAS methods rely on the weight-sharing idea. In this approach, a single overparameterized supernetwork (also known as the one-shot model) is defined. A supernetwork is a very large Directed Acyclic Graph (DAG) whose subgraphs are different candidate neural networks. Thus, in a supernetwork, the weights are shared among a large number of different sub-architectures that have edges in common, each of which is considered as a path within the supernet. The essential idea is to train one supernetwork that spans many options for the final design rather than generating and training thousands of networks independently. In addition to the learned parameters, a set of architecture parameters are learnt to depict preference for one module over another. Such methods reduce the required computational resources to only a few GPU days. More recent works further combine this weight-sharing paradigm, with a continuous relaxation of the search space, which enables the use of gradient-based optimization methods. These approaches are generally referred to as differentiable NAS and have proven very efficient in exploring the search space of ne

    Read more →
  • Google AI Studio

    Google AI Studio

    Google AI Studio is a web-based integrated development environment developed by Google for prototyping applications using generative AI models. Released in December 2023 alongside the Gemini API, the platform provides access to Google's Gemini family of models and related tools for image, video, and audio generation. The service targets both developers and non-technical users for testing prompts and generating code for the Gemini API. == History == Google launched AI Studio on December 13, 2023, as the successor to Google MakerSuite. MakerSuite, introduced at Google I/O in May 2023, had provided similar functionality for Google's PaLM language models. The AI Studio was launched alongside the public release of the Gemini API. == Features == AI Studio's interface consists of a central prompt area and a settings panel for model selection and parameter adjustment. The platform supports chat prompts for multi-turn conversations and includes system instructions for defining model behavior, tone, or specific rules. Users can employ zero-shot and few-shot prompting techniques to guide the model's output format. The platform processes various media types including video, audio, and documents, and can generate images through Imagen models, videos through Veo models, and audio through text-to-speech functionality. Additional tools include real-time streaming for screen sharing and live analysis, code execution in a sandboxed Python environment, grounding with Google Search for current information, URL context for analyzing specific web pages, and a thinking mode for complex reasoning tasks. == Available models == The platform provides access to several Google AI models including the Gemini language models, Imagen for image generation, Veo for video generation, LearnLM for educational applications, and Gemma, Google's open-source model family. == Privacy and data usage == Google AI Studio's data handling differs between free and paid users. For free tier users, Google uses submitted prompts, uploaded files, and generated responses to improve its products and services, with human reviewers potentially reading and annotating the data after disconnection from user accounts. Google advises against submitting sensitive information on the free tier. Users who enable Google Cloud Billing are considered paid service users, and their data is not used for product improvement. Data is processed according to Google's Data Processing Addendum and retained temporarily for abuse monitoring. == Availability == The platform is available at no cost, with API usage subject to a free tier with daily and per-minute rate limits. Access is restricted to users aged 18 and older in specific countries and territories. The service was initially unavailable in the United Kingdom and European Economic Area due to regulatory concerns, which drew user complaints. == Reception == Reviews have noted the platform's accessibility and integration with Gemini models, with features such as real-time screen sharing and large context windows cited as notable capabilities. However, reviewers have raised concerns about the privacy implications for free tier users, whose data is used for model training. Some users have reported inconsistent performance with features like screen streaming and issues with folder uploads for large datasets. The initial geographic restrictions were a point of criticism among developers in affected regions.

    Read more →