AI Data Center Map

AI Data Center Map — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

• 

  Pocket (service)

  Pocket, formerly known as Read It Later, was a social bookmarking service for storing, sharing and discovering web bookmarks, first released in 2007. Mozilla, the developer of Pocket, announced in May 2025 that it was discontinuing the service and would shut it down in July of that year. == History == Pocket was introduced in August 2007 as a Mozilla Firefox browser extension named Read It Later by Nathan (Nate) Weiner. Once his product was used by millions of people, he moved his office to Silicon Valley and four other people joined the Read It Later team. Weiner's intention was for the application to be like a TiVo directory for web content and to give users access to that content on any device. Read It Later obtained venture capital investments of US$2.5 million in 2011 and $5.0 million in 2012. The 2011 funding came from Foundation Capital, Baseline Ventures, Google Ventures, Founder Collective and unnamed angel investors. The company rejected an acquisition offer by Evernote after showing concerns that Evernote intended to shut down the Read It Later service and amalgamate its functionality into Evernote's main service. Initially, the Read It Later app was available in a free version and a paid version that included additional features. After the rebranding to Pocket, all paid features were made available in a free and advertisement-free app. In May 2014, a paid subscription service called Pocket Premium was introduced, adding server-side storage of articles and more powerful search tools. In June 2015, Pocket was included in Firefox, via a toolbar button and link to a user's Pocket list in the bookmark's menu. The integration was controversial, as users displayed concerns for the direct integration of a proprietary service into an open source application, and that it could not be completely disabled without editing advanced settings, unlike other third-party extensions. A Mozilla spokesperson stated that the feature was meant to leverage the service's popularity among Firefox users and clarified that all code related to the integration was open source. The spokesperson added that "[Mozilla had] gotten lots of positive feedback about the integration from users". On February 27, 2017, Pocket announced that it had been acquired by Mozilla Corporation, the commercial arm of Firefox's non-profit development group. Mozilla staff stated that Pocket would continue to operate as an independent subsidiary but that it would be leveraged as part of an ongoing "Context Graph" project. There were plans to open-source the server-side code of Pocket, though only parts of the project had been open-sourced as of 2024. On May 22, 2025, Mozilla announced that it would shut down Pocket on July 8, 2025. Exports of user data would be available until October 8, 2025, when accounts would be deleted. The email newsletter Pocket Hits was rebranded as Ten Tabs on June 12 as part of the closure, with it being changed to release only on weekdays. == Functions == The application allows the user to save an article or web page to remote servers for later reading. The article is sent to the user's Pocket list (synced to all of their devices) for offline reading. Pocket makes the article more readable by removing clutter and enabling the user to add tags and adjust text settings. == User base == The application had 17 million users and 1 billion saves, as of September 2015. Pocket was listed among Time magazine's 50 Best Android Applications for 2013. == Reception == Kent German of CNET said that "Read It Later is oh so incredibly useful for saving all the articles and news stories I find while commuting or waiting in line." Erez Zukerman of PC World said that supporting the developer is enough reason to buy what he deemed a "handy app". Bill Barol of Forbes said that although Read It Later works less well than Instapaper, "it makes my beloved Instapaper look and feel a little stodgy." In 2015, Pocket was awarded a Material Design Award for Adaptive Layout by Google for their Android application.

  Read more →
• 

  MultiNet

  Multilayered extended semantic networks (MultiNets) are both a knowledge representation paradigm and a language for meaning representation of natural language expressions that has been developed by Prof. Dr. Hermann Helbig on the basis of earlier Semantic Networks. It is used in a question-answering application for German called InSicht. It is also used to create a tutoring application developed by the university of University of Hagen to teach MultiNet to knowledge engineers. MultiNet is claimed to be one of the most comprehensive and thoroughly described knowledge representation systems. It specifies conceptual structures by means of about 140 predefined relations and functions, which are systematically characterized and underpinned by a formal axiomatic apparatus. Apart from their relational connections, the concepts are embedded in a multidimensional space of layered attributes and their values. Another characteristic of MultiNet distinguishing it from simple semantic networks is the possibility to encapsulate whole partial networks and represent the resulting conceptual capsule as a node of higher order, which itself can be an argument of relations and functions. MultiNet has been used in practical NLP applications such as natural language interfaces to the Internet or question answering systems over large semantically annotated corpora with millions of sentences. MultiNet is also a cornerstone of the commercially available search engine SEMPRIA-Search, where it is used for the description of the computational lexicon and the background knowledge, for the syntactic-semantic analysis, for logical answer finding, as well as for the generation of natural language answers. MultiNet is supported by a set of software tools and has been used to build large semantically based computational lexicons. The tools include a semantic interpreter WOCADI, which translates natural language expressions (phrases, sentences, texts) into formal MultiNet expressions, a workbench MWR+ for the knowledge engineer (comprising modules for automatic knowledge acquisition and reasoning), and a workbench LIA+ for the computer lexicographer supporting the creation of large semantically based computational lexica.

  Read more →
• 

  Learning Applied to Ground Vehicles

  The Learning Applied to Ground Vehicles (LAGR) program, which ran from 2004 until 2008, had the goal of accelerating progress in autonomous, perception-based, off-road navigation in robotic unmanned ground vehicles (UGVs). LAGR was funded by DARPA, a research agency of the United States Department of Defense. == History and background == While mobile robots had been in existence since the 1960s, (e.g. Shakey), progress in creating robots that could navigate on their own, outdoors, off-road, on irregular, obstacle-rich terrain had been slow. In fact, no clear metrics were in place to measure progress. A baseline understanding of off-road capabilities began to emerge with the DARPA PerceptOR program in which independent research teams fielded robotic vehicles in unrehearsed Government tests that measured average speed and number of required operator interventions over a fixed course over widely spaced waypoints. These tests exposed the extreme challenges of off-road navigation. While the PerceptOR vehicles were equipped with sensors and algorithms that were state-of-the-art for the beginning of the 21st century, the limited range of their perception technology caused them to become trapped in natural cul-de-sacs. Furthermore, their reliance on pre-scripted behaviors did not allow them to adapt to unexpected circumstances. The overall result was that except for essentially open terrain with minimal obstacles, or along dirt roads, the PerceptOR vehicles were unable navigate without numerous, repeated operator intervention. The LAGR program was designed to build on the methodology started in PerceptOR while seeking to overcome the technical challenges exposed by the PerceptOR tests. == LAGR goals == The principal goal of LAGR was to accelerate progress in off navigation of UGVs. Additional, synergistic goals included (1) establishing benchmarking methodology for measuring progress for autonomous robots operating in unstructured environments, (2) advancing machine vision and thus enabling long-range perception, and (3) increasing the number of institutions and individuals who were able to contribute to forefront UGV research. == Structure and rationale of the LAGR program == The LAGR program was designed to focus on developing new science for robot perception and control rather than on new hardware. Thus, it was decided to create a fleet of identical, relatively simple robots that would be supplied to the LAGR researchers, who were members of competitive teams, freeing them to concentrate on algorithm development. The teams were each given two robots of the standard design. They developed new software on these robots, and then sent the code to a government test team that then tested that code on Government robots at various test courses. These courses were located throughout the US and were not previously known to the teams. In this way, the code from all teams could be tested in essentially identical circumstances. After an initial startup period, the code development/test cycle was repeated about once every month. The standard robot was designed and built by the Carnegie Mellon University National Robotics Engineering Center (CMU NREC). The vehicles’ computers were preloaded with a modular “Baseline” perception and navigation system that was essentially the same system that CMU NREC had created for the PerceptOR program and was considered to represent the state-of-the-art at the inception of LAGR. The modular nature of the Baseline system allowed the researchers to replace parts of the Baseline code with their own modules and still have a complete working system without having to create an entire navigation system from scratch. Thus, for example, they were able to compare the performance of their own obstacle detection module with that of the Baseline code, while holding everything else fixed. The Baseline code also served as a fixed reference – in any environment and at any time in the program, teams’ code could be compared to the Baseline code. This rapid cycle gave the Government team and the performer teams quick feedback and allowed the Government team to design test courses that challenged the performers in specific perception tasks and whose difficulty was likely to challenge, but not overwhelm, the performers’ current capabilities. Teams were not required to submit new code for every test, but usually did. Despite this leeway, some teams found the rapid test cycle distracting to their long term progress and would have preferred a longer interval between tests. === Phase II === To advance to Phase II, each team had to modify the Baseline code so that on the final 3 tests of Phase I of the government tests, robots running the team's code averaged at least 10% faster than a vehicle running the original Baseline code. This rather modest “Go/ No Go” metric was chosen to allow teams to choose risky, but promising approaches that might not be fully developed in the first 18 months of the program. All 8 teams achieved this metric, with some scoring more twice the speed of the Baseline on the later tests which was the objective for Phase II. Note that the Phase I Go / No Go metric was such that teams were not in completion with each other for a limited number of slots on Phase II: any number of teams, from eight to zero could make the grade. This strategy by DARPA was to designed to encourage cooperation and even code sharing among the teams. == The LAGR teams == Eight teams were selected as performers in Phase I, the first 18 months of LAGR. The teams were from Applied Perception (Principal Investigator [PI] Mark Ollis), Georgia Tech (PI Tucker Balch), Jet Propulsion Laboratory (PI Larry Matthies), Net-Scale Technologies (PI Urs Muller), NIST (PI James Albus), Stanford University (PI Sebastian Thrun), SRI International (PI Robert Bolles), and University of Pennsylvania (PI Daniel Lee). The Stanford team resigned at the end of Phase I to focus its efforts on the DARPA Grand Challenge; it was replaced by a team from the University of Colorado, Boulder (PI Greg Grudic). Also in Phase II, the NIST team suspended its participation in the competition and instead concentrated on assembling the best software elements from each team into a single system. Roger Bostelman became PI of that effort. == The LAGR vehicle == The LAGR vehicle, which was about the size of a supermarket shopping cart, was designed to be simple to control. (A companion DARPA program, Learning Locomotion, addressed complex motor control.) It was battery powered and had two independently driven wheelchair motors in the front, and two caster wheels in the rear. When the front wheels were rotated in the same direction the robot was driven either forward or reverse. When these wheels were driven in opposite directions, the robot turned. The ~ $30,000 cost of the LAGR vehicle meant that a fleet could be built and distributed to a number of teams expanding on the field of researchers who had traditionally participated in DARPA robotics programs. The vehicle's top speed of about 3 miles/ hour and relatively modest weight of ~100 kg meant that it posed a much reduced safety hazard compared to vehicles used in previous programs in unmanned ground vehicles and thus further reduced the budget required for each team to manage its robot. Nevertheless, the LAGR vehicles were sophisticated machines. Their sensor suite included 2 pairs of stereo cameras, an accelerometer, a bumper sensor, wheel encoders, and a GPS. The vehicle also had three computers that were user-programmable. == Scientific results == A cornerstone of the program was incorporation of learned behaviors in the robots. In addition, the program used passive optical systems to accomplish long-range scene analysis. The difficulty of testing UGV navigation in unstructured, off-road environments made accurate, objective measurement of progress a challenging task. While no absolute measure of performance had been defined in LAGR, the relative comparison of a team's code to that of the Baseline code on a given course demonstrated whether progress was being made in that environment. By the conclusion of the program, testing showed that many of the performers had attained leaps in performance. In particular, average autonomous speeds were increased by factor of 3 and useful visual perception was extended to ranges as far as 100 meters. While LAGR did succeed in extending the useful range of visual perception, this was primarily done by either pixel or patch-based color or texture analysis. Object recognition was not directly addressed. Even though the LAGR vehicle had a WAAS GPS, its position was never determined down to the width of the vehicle, so it was hard for the systems to re-use obstacle maps of areas the robots had previously traversed since the GPS continually drifted. The drift was especially severe if there was a forest canopy. A few teams developed visual odometry algorithms that essentially eliminated this drift.

  Read more →
• 

  PyTorch

  PyTorch is an open-source deep learning library, originally developed by Meta Platforms and currently developed with support from the Linux Foundation. The successor to Torch, PyTorch provides a high-level API that builds upon optimised, low-level implementations of deep learning algorithms and architectures, such as the Transformer, or SGD. Notably, this API simplifies model training and inference to a few lines of code. PyTorch allows for automatic parallelization of training and, internally, implements CUDA bindings that speed training further by leveraging GPU resources. PyTorch utilises the tensor as a fundamental data type, similarly to NumPy. Training is facilitated by a reversed automatic differentiation system, Autograd, that constructs a directed acyclic graph of the operations (and their arguments) executed by a model during its forward pass. With a loss, backpropagation is then undertaken. As of 2025, PyTorch remains one of the most popular deep learning libraries, alongside others such as TensorFlow and Keras. It can be installed using Anaconda package managers. A number of commercial deep learning architectures are built on top of PyTorch, including ChatGPT, Tesla Autopilot, Uber's Pyro, and Hugging Face's Transformers. == History == In 2001, Torch was written and released under a GPL. It was a machine-learning library written in C++ and CUDA, supporting methods including neural networks, support vector machines (SVM), hidden Markov models, etc. Around 2010, it was rewritten by Ronan Collobert, Clement Farabet and Koray Kavuckuoglu. This was known as Torch7 or LuaTorch. This was written so that the backend was in C and the frontend was in Lua. In mid-2016, some developers refactored it to decouple the frontend and the backend, with strong influence from torch-autograd and Chainer. In turn, torch-autograd was influenced by HIPS/autograd. Development on Torch7 ceased in 2018 and was subsumed by the PyTorch project. Meta (formerly known as Facebook) operates both PyTorch and Convolutional Architecture for Fast Feature Embedding (Caffe2), but models defined by the two frameworks were mutually incompatible. The Open Neural Network Exchange (ONNX) project was created by Meta and Microsoft in September 2017 to decouple deep learning frameworks from hardware-specific runtimes, allowing models to be converted between frameworks and optimized for execution providers like NVIDIA’s TensorRT. Caffe2 was merged into PyTorch at the end of March 2018. In September 2022, Meta announced that PyTorch would be governed by the independent PyTorch Foundation, a newly created subsidiary of the Linux Foundation. PyTorch 2.0 was released on 15 March 2023, introducing TorchDynamo, a Python-level compiler that makes code run up to two times faster, along with significant improvements in training and inference performance across major cloud platforms. == PyTorch tensors == PyTorch defines a class called Tensor (torch.Tensor) to store and operate on homogeneous multidimensional rectangular arrays of numbers. PyTorch supports various sub-types of multi-dimensional arrays, or Tensors. PyTorch Tensors are similar to NumPy Arrays, but can also be operated on by a CUDA-capable NVIDIA GPU. PyTorch has also been developing support for other GPU platforms, for example, AMD's ROCm and Apple's Metal Framework. == PyTorch neural networks == PyTorch defines a module called nn (torch.nn) to describe neural networks and to support training. This module offers a comprehensive collection of building blocks for neural networks, including various layers and activation functions, enabling the construction of complex models. Networks are built by inheriting from the torch.nn module and defining the sequence of operations in the forward() function. == PyTorch Serialized File Format == Pytorch can save and load models using its own file format, which is a ZIP64 archive containing the model weights in a Python pickle file, and other information such as the byte order. The file extensions .pt and .pth are commonly used for these files. == Example == The following program shows the low-level functionality of the library with a simple example. The following code block defines a neural network with linear layers using the nn module.

  Read more →
• 

  BERT (language model)

  Bidirectional encoder representations from transformers (BERT) is a language model introduced in October 2018 by researchers at Google. It learns to represent text as a sequence of vectors using self-supervised learning. It uses the encoder-only transformer architecture. BERT dramatically improved the state of the art for large language models. As of 2020, BERT is a ubiquitous baseline in natural language processing (NLP) experiments. BERT is trained by masked token prediction and next sentence prediction. With this training, BERT learns contextual, latent representations of tokens in their context, similar to ELMo and GPT-2. It found applications for many natural language processing tasks, such as coreference resolution and polysemy resolution. It improved on ELMo and spawned the study of "BERTology", which attempts to interpret what is learned by BERT. BERT was originally implemented in the English language at two model sizes, BERTBASE (110 million parameters) and BERTLARGE (340 million parameters). Both were trained on the Toronto BookCorpus (800M words) and English Wikipedia (2,500M words). The weights were released on GitHub. On March 11, 2020, 24 smaller models were released, the smallest being BERTTINY with just 4 million parameters. == Architecture == BERT is an "encoder-only" transformer architecture. At a high level, BERT consists of 4 modules: Tokenizer: This module converts a piece of English text into a sequence of integers ("tokens"). Embedding: This module converts the sequence of tokens into an array of real-valued vectors representing the tokens. It represents the conversion of discrete token types into a lower-dimensional Euclidean space. Encoder: a stack of Transformer blocks with self-attention, but without causal masking. Task head: This module converts the final representation vectors into one-shot encoded tokens again by producing a predicted probability distribution over the token types. It can be viewed as a simple decoder, decoding the latent representation into token types, or as an "un-embedding layer". The task head is necessary for pre-training, but it is often unnecessary for so-called "downstream tasks," such as question answering or sentiment classification. Instead, one removes the task head and replaces it with a newly initialized module suited for the task, and finetune the new module. The latent vector representation of the model is directly fed into this new module, allowing for sample-efficient transfer learning. === Embedding === This section describes the embedding used by BERTBASE. The other one, BERTLARGE, is similar, just larger. The tokenizer of BERT is WordPiece, which is a sub-word strategy like byte-pair encoding. Its vocabulary size is 30,000, and any token not appearing in its vocabulary is replaced by [UNK] ("unknown"). The first layer is the embedding layer, which contains three components: token type embeddings, position embeddings, and segment type embeddings. Token type: The token type is a standard embedding layer, translating a one-hot vector into a dense vector based on its token type. Position: The position embeddings are based on a token's position in the sequence. BERT uses absolute position embeddings, where each position in a sequence is mapped to a real-valued vector. Each dimension of the vector consists of a sinusoidal function that takes the position in the sequence as input. Segment type: Using a vocabulary of just 0 or 1, this embedding layer produces a dense vector based on whether the token belongs to the first or second text segment in that input. In other words, type-1 tokens are all tokens that appear after the [SEP] special token. All prior tokens are type-0. The three embedding vectors are added together representing the initial token representation as a function of these three pieces of information. After embedding, the vector representation is normalized using a LayerNorm operation, outputting a 768-dimensional vector for each input token. After this, the representation vectors are passed forward through 12 Transformer encoder blocks, and are decoded back to 30,000-dimensional vocabulary space using a basic affine transformation layer. === Architectural family === The encoder stack of BERT has 2 free parameters: L {\displaystyle L} , the number of layers, and H {\displaystyle H} , the hidden size. There are always H / 64 {\displaystyle H/64} self-attention heads, and the feed-forward/filter size is always 4 H {\displaystyle 4H} . By varying these two numbers, one obtains an entire family of BERT models. For BERT: the feed-forward size and filter size are synonymous. Both of them denote the number of dimensions in the middle layer of the feed-forward network. the hidden size and embedding size are synonymous. Both of them denote the number of real numbers used to represent a token. The notation for encoder stack is written as L/H. For example, BERTBASE is written as 12L/768H, BERTLARGE as 24L/1024H, and BERTTINY as 2L/128H. == Training == === Pre-training === BERT was pre-trained simultaneously on two tasks: Masked language modeling (MLM): In this task, BERT ingests a sequence of words, where one word may be randomly changed ("masked"), and BERT tries to predict the original words that had been changed. For example, in the sentence "The cat sat on the [MASK]," BERT would need to predict "mat." This helps BERT learn bidirectional context, meaning it understands the relationships between words not just from left to right or right to left but from both directions at the same time. Next sentence prediction (NSP): In this task, BERT is trained to predict whether one sentence logically follows another. For example, given two sentences, "The cat sat on the mat" and "It was a sunny day", BERT has to decide if the second sentence is a valid continuation of the first one. This helps BERT understand relationships between sentences, which is important for tasks like question answering or document classification. ==== Masked language modeling ==== In masked language modeling, 15% of tokens would be randomly selected for masked-prediction task, and the training objective was to predict the masked token given its context. In more detail, the selected token is: replaced with a [MASK] token with probability 80%, replaced with a random word token with probability 10%, not replaced with probability 10%. The reason not all selected tokens are masked is to avoid the dataset shift problem. The dataset shift problem arises when the distribution of inputs seen during training differs significantly from the distribution encountered during inference. A trained BERT model might be applied to word representation (like Word2Vec), where it would be run over sentences not containing any [MASK] tokens. It is later found that more diverse training objectives are generally better. As an illustrative example, consider the sentence "my dog is cute". It would first be divided into tokens like "my1 dog2 is3 cute4". Then a random token in the sentence would be picked. Let it be the 4th one "cute4". Next, there would be three possibilities: with probability 80%, the chosen token is masked, resulting in "my1 dog2 is3 [MASK]4"; with probability 10%, the chosen token is replaced by a uniformly sampled random token, such as "happy", resulting in "my1 dog2 is3 happy4"; with probability 10%, nothing is done, resulting in "my1 dog2 is3 cute4". After processing the input text, the model's 4th output vector is passed to its decoder layer, which outputs a probability distribution over its 30,000-dimensional vocabulary space. ==== Next sentence prediction ==== Given two sentences, the model predicts if they appear sequentially in the training corpus, outputting either [IsNext] or [NotNext]. During training, the algorithm sometimes samples two sentences from a single continuous span in the training corpus, while at other times, it samples two sentences from two discontinuous spans. The first sentence starts with a special token, [CLS] (for "classify"). The two sentences are separated by another special token, [SEP] (for "separate"). After processing the two sentences, the final vector for the [CLS] token is passed to a linear layer for binary classification into [IsNext] and [NotNext]. For example: Given "[CLS] my dog is cute [SEP] he likes playing [SEP]", the model should predict [IsNext]. Given "[CLS] my dog is cute [SEP] how do magnets work [SEP]", the model should predict [NotNext]. === Fine-tuning === BERT is meant as a general pretrained model for various applications in natural language processing. That is, after pre-training, BERT can be fine-tuned with fewer resources on smaller datasets to optimize its performance on specific tasks such as natural language inference and text classification, and sequence-to-sequence-based language generation tasks such as question answering and conversational response generation. The original BERT paper published results demonstrating that a small amount of fine

  Read more →
• 

  Department of Defense Directive 3000.09

  Department of Defense Directive 3000.09 (DODD 3000.09), titled Autonomy in Weapon Systems, is the current U.S. military policy on autonomous weapons. It states: "Autonomous and semi-autonomous weapon systems will be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force." == History == Then-Deputy Secretary of Defense Ashton Carter issued DOD's policy on autonomy in weapons systems, Department of Defense Directive (DODD) 3000.09, in November 2012. DOD updated the directive in January 2023. In February 2023, the US issued a related foreign policy proposal, Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy. == Definitions == There is no agreed definition of lethal autonomous weapon systems that is used in international fora. However, DODD 3000.09 provides definitions for different categories of autonomous weapon systems for the purposes of the U.S. military. These definitions are principally grounded in the role of the human operator with regard to target selection and engagement decisions, rather than in the technological sophistication of the weapon system. DODD 3000.09 defines LAWS as "weapon system[s] that, once activated, can select and engage targets without further intervention by a human operator." This concept of autonomy is also known as "human out of the loop" or "full autonomy." The directive contrasts LAWS with human-supervised, or "human on the loop," autonomous weapon systems, in which operators have the ability to monitor and halt a weapon's target engagement. Another category is semi-autonomous, or "human in the loop," weapon systems that "only engage individual targets or specific target groups that have been selected by a human operator." Semi-autonomous weapons include so-called "fire and forget" weapons, such as certain types of guided missiles, that deliver effects to human-identified targets using autonomous functions. The directive does not apply to autonomous or semi-autonomous cyberspace capabilities; unarmed platforms; unguided munitions; munitions manually guided by the operator (e.g., laser- or wire-guided munitions); mines; unexploded explosive ordnance; or autonomous or semi-autonomous systems that are not weapon systems, nor subject them to its guidelines. == Role of human operator == DODD 3000.09 requires that all systems, including LAWS, be designed to "allow commanders and operators to exercise appropriate levels of human judgment over the use of force." As noted in an August 2018 U.S. government white paper, "'appropriate' is a flexible term that reflects the fact that there is not a fixed, one-size-fits-all level of human judgment that should be applied to every context. What is 'appropriate' can differ across weapon systems, domains of warfare, types of warfare, operational contexts, and even across different functions in a weapon system." Furthermore, "human judgment over the use of force" does not require manual human "control" of the weapon system, as is often reported, but rather broader human involvement in decisions about how, when, where, and why the weapon will be employed. This includes a human determination that the weapon will be used "with appropriate care and in accordance with the law of war, applicable treaties, weapon system safety rules, and applicable rules of engagement." To aid this determination, DODD 3000.09 requires that "[a]dequate training, [tactics, techniques, and procedures], and doctrine are available, periodically reviewed, and used by system operators and commanders to understand the functioning, capabilities, and limitations of the system's autonomy in realistic operational conditions." The directive also requires that the weapon's human-machine interface be "readily understandable to trained operators" so they can make informed decisions regarding the weapon's use. == Weapons review process == DODD 3000.09 requires that the software and hardware of covered semi-autonomous and autonomous weapon systems, be tested and evaluated to ensure they:Function as anticipated in realistic operational environments against adaptive adversaries taking realistic and practicable countermeasures, [and] complete engagements within a timeframe and geographic area, as well as other relevant environmental and operational constraints, consistent with commander and operator intentions. If unable to do so, the systems will terminate the engagement or obtain additional operator input before continuing the engagement.Systems must also be "sufficiently robust to minimize the probability and consequences of failures." Any changes to the system's operating state—for example, due to machine learning—would require the system to go through testing and evaluation again to ensure that it has retained its safety features and ability to operate as intended. The directive also notes that "the use of AI capabilities in autonomous or semi-autonomous systems will be consistent with the DOD AI Ethical Principles." In addition to the standard weapons review process, a secondary senior-level review is required for covered autonomous and semi-autonomous systems. This review requires the Under Secretary of Defense for Policy (USD[P]), the vice chairman of the Joint Chiefs of Staff (VCJCS), and the Under Secretary of Defense for Research and Engineering (USD[R&E]) to approve the system before formal development. USD(P), VCJCS, and the Under Secretary of Defense for Acquisition and Sustainment (USD[A&S]) must then approve the system before fielding. In the event of "urgent military need," this senior-level review may be waived by the Deputy Secretary of Defense. DODD 3000.09 additionally establishes the Autonomous Weapon System Working Group—composed of representatives of USD(P); USD(R&E); USD(A&S); DOD General Counsel; the Chief Digital and AI Officer; the Director, Operational Test and Evaluation; and the chairman of the Joint Chiefs of Staff—to support and advise the senior-level review process. == Congressional notification == Per Section 251 of the FY2024 National Defense Authorization Act (NDAA; Pub. L. 118–31 (text) (PDF)), the Secretary of Defense is to notify the defense committees of any changes to DODD 3000.09 within 30 days. The Secretary is directed to provide a description of the modification and an explanation of the reasons for the modification. Section 1066 of the FY2025 NDAA (Pub. L. 118–159 (text) (PDF)) additionally requires the Secretary to "submit to the congressional defense committees a comprehensive report on the approval and deployment of lethal autonomous weapon systems by the United States," annually through December 31, 2029. Section 1061 of the FY2026 NDAA (P.L. Pub. L. 119–60 (menu; GPO has not yet published law)) amends the U.S. Code to require congressional notification of any waiver issued under DODD 3000.09. == AI safety == The second revision of DoDD 3000.09, effective January 25, 2023, requires that "The DoD will design and engineer AI capabilities to fulfill their intended functions while possessing the ability to detect and avoid unintended consequences, and the ability to disengage or deactivate deployed systems that demonstrate unintended behavior." == Criticism == As noted in the Bulletin of the Atomic Scientists, the policy requires that autonomous weapon systems that kill people or use kinetic force, selecting and engaging targets without further human intervention, be certified as compliant with "appropriate levels" and other standards, not that such weapon systems cannot meet these standards and are therefore forbidden. "Semi-autonomous" hunter-killers that autonomously identify and attack targets do not require certification.

  Read more →
• 

  Botler AI

  Botler AI is a Montreal-based Canadian Artificial Intelligence company that helps users navigate the legal system. Launched in 2017 by Amir Morv and Ritika Dutt, Botler offers a free online tool which provides users who are unaware of their legal rights with information and guidance. Botler is known for its role in unveiling misconduct in the Government of Canada's procurement practices. Botler's findings have prompted numerous investigations, including by the Royal Canadian Mounted Police. == History == Botler's first AI was trained on over 300,000 U.S. and Canadian legal documents to help individuals identify and enforce their legal rights, without fear of judgment. Launched during the height of the #MeToo movement, the tool initially focused on sexual harassment with a goal of creating "a general artificial intelligence that would help the average person with any legal issue." === Department of Justice Canada === In 2020, Botler launched an expanded misconduct detection system in the form of an anonymous chatbot which provided users with an explanation of the law and relevant resources. In March 2021, the Minister of Justice and Attorney General of Canada announced the Government of Canada's support for Botler AI to assist complainants of sexual harassment in the workplace. The initiative, entitled Botler for Citizens and implemented with the support of the Department of Justice Canada, established an Artificial Intelligence-powered hybrid legal service delivery model. == Notable cases == On October 4, 2023, the RCMP confirmed to The Globe and Mail that they "are investigating a file referred from the CBSA (Canada Border Services Agency) that is based on allegations brought to their attention by Botler". In 2019, GCStrategies's managing partner, Kristian Firth, reached out to Botler on behalf of his client, the CBSA, to solicit their misconduct detection chatbot. After interactions with GCStrategies, Dalian Enterprises and Coradix Technology Consulting, the three main contractors involved in developing the controversial ArriveCAN app, Dutt and Morv alerted the CBSA to questionable contracting practices in federal government procurement in September, 2021, and again in November, 2022. In response to Botler's November 2022 report, the CBSA launched an internal review and referred the matter to the RCMP. During testimony before a parliamentary committee, the CBSA's President stated that the CBSA investigation to date has raised some concerns and shows "that there was a pattern of persistent collaboration between certain officials and GCStrategies... to circumvent or ignore certain established processes and roles and responsibilities". The Auditor General of Canada, which extended its study into ArriveCAN following the Botler revelations, found that GCStrategies was directly involved in setting narrow terms for a request for proposal for a $25-million government contract it ultimately won. The firm, which has just two employees, charges the government a commission of between 15 per cent and 30 per cent of each contract's value. The Office of the Procurement Ombudsman of Canada found "numerous examples" where GCStrategies "had simply copied and pasted" the required work experience to meet contracting requirements. To date, more than a dozen probes have been launched into the matter, including by the government, parliamentary committees, independent watchdogs and law-enforcement agencies. On April 17, 2024, GCStrategies' Firth was the first person summoned in over a century to answer questions before Members of Parliament in the House of Commons. During his appearance, Firth testified that the RCMP had raided "my property to obtain electronic goods surrounding the Botler allegations". === Government of Canada Reforms === One day after The Globe reported that the RCMP is investigating allegations of misconduct, the federal government responded by announcing new guidelines from the Treasury Board of Canada aimed at cutting back on the use of private consultants and that outsourcing contracts were under examination. Public Services and Procurement Canada (PSPC) invalidated and replaced all master level user agreements with government client departments in November 2023. The agreements set out the conditions for access to select Professional Services methods of supply which are used for outsourcing. In March 2024, PSPC announced its suspension of the respective security statuses of GCStrategies, Dalian and Coradix, barring them from participating in all federal procurements. Records show that the total value of contracts awarded to the three companies amounts to more than $1 Billion.

  Read more →
• 

  Visual hierarchy

  Visual hierarchy, in Gestalt psychology, describes how particular elements in a visual field stand out more than others in a pattern, creating a perceived order of importance. Although it can occur naturally, the term is most often used in design—especially graphic design and cartography—where elements are arranged to appear more important than others. This order is created by the visual contrast between forms in a field of perception. Objects with highest contrast to their surroundings are recognized first by the human mind. == Evidence == There is some scientific evidence for visual hierarchy using eye tracking. For example, one study found that when people agree that a graphic design is good, they exhibit more similar eye movements; measured by the Fréchet distance. == Theory == The concept of visual hierarchy is based in Gestalt psychological theory, an early 20th-century German theory that proposes that the human brain has innate organizing tendencies that “structure individual elements, shapes or forms into a coherent, organized whole,” especially when processing visual information. The German word Gestalt translates into “form,” “pattern,” or “shape” in English. When an element in a visual field disconnects from the ‘whole’ created by the brain's perceptual organization, it “stands out” to the viewer. The shapes that disconnect most severely from their surroundings stand out the most. This is commonly encapsulated as the Von Restorff effect, which states that isolation attracts attention. === Physical characteristics === The brain distinguishes objects based on differences in their physical appearances. These characteristics fall into four categories: color, size, alignment, and character. Each type of contrast can be used to construct a visual hierarchy. The same characteristics are also sometimes categorized (especially among cartographers) according to the visual variables of Jacques Bertin. Color encompasses the hue, saturation, value, and perceived texture of forms. Dark figures will stand out on a light background, light figures will stand out on a dark background, brightly colored figures will stand out on a muted background, and so on. The fluorescent colors used for tennis balls and other sports equipment is intended to make them instantly stand out against almost any natural visual field. Size has a strong influence on visual hierarchy. Large elements typically attract attention, provided that they can be recognized as figures. Alignment is the arrangement of forms relative to one another. For example, items in the upper left corner of a page are often seen first (at least for those readers accustomed to western languages), the center of the field has prominence. Negative space can also be employed: a figure isolated among large amounts of white space will stand out more than one amid other figures. Character includes several kinds of contrasts based on shape. For example, complex patterns attract more attention than simple or predictable patterns, intricate shapes attract more attention than generalized ones. Even large-scale patterns can attract attention if they contrast with the pattern in the remainder of the visual field. Camouflage is an example of eliminating contrast in character in color and/or character specifically to reduce visual hierarchy. The "squint test" is often suggested as a simple, if unscientific, method to evaluate the visual hierarchy of a graphical product like a map or web page. When viewed out of focus (or from a great distance), the viewer is not distracted by details, but can only see overall (gestalt) patterns such as visual hierarchy. All of the above patterns, except some aspects of character, are recognizable by this method. == Application == Visual hierarchy is an important concept in the field of graphic design, a field that specializes in visual organization. Designers attempt to control visual hierarchy to guide the eye to information in a specific order for a specific purpose. One could compare visual hierarchy in graphic design to grammatical structure in writing in terms of the importance of each principle to these fields. === Cartography === In cartographic design, visual hierarchy is used to emphasize certain important features on a map over less important features. Typically, a map has a purpose that dictates a conceptual hierarchy of what should be more or less important, so one of the goals of the choice of map symbols is to match the visual hierarchy to the conceptual hierarchy. The Visual hierarchy of a map may apply to individual geographic features (such as making a single country stand out), to map layers of related features (e.g., making lakes stand out more than roads), and to the entire layout of map and non-map elements (e.g., making the title look more important than the scale bar). Like the main map elements, such features have weight, and the properties that apply to visual hierarchy of map layers also apply to other elements on the page. Size and alignment are the two main determinants of the visual hierarchy for these features. Cartographers often utilize principles of negative space and figure-ground contrast to design an appropriate visual hierarchy by employing contrast between unused space and layout features. === User experience design and behavioral design === In user experience design and behavioural design, such as web design, visual hierarchy is used to prioritize navigational structures and content, so that audiences focus on elements that facilitate system usage, or increases the chance that they notice content that contains psychological nudges. Color is one of many factors used in the design of a visual hierarchy, and a key factor due to the high salience of color perception.

  Read more →
• 

  Projection-slice theorem

  In mathematics, the projection-slice theorem, central slice theorem or Fourier slice theorem in two dimensions states that the results of the following two calculations are equal: Take a two-dimensional function f(r), project (e.g. using the Radon transform) it onto a (one-dimensional) line, and do a Fourier transform of that projection. Take that same function, but do a two-dimensional Fourier transform first, and then slice the function through its origin, parallel to the projection line. In operator terms, if F1 and F2 are the 1- and 2-dimensional Fourier transform operators mentioned above, P1 is the projection operator (which projects a 2-D function onto a 1-D line), S1 is a slice operator (which extracts a 1-D central slice from a function), then F 1 P 1 = S 1 F 2 . {\displaystyle F_{1}P_{1}=S_{1}F_{2}.} This idea can be extended to higher dimensions. This theorem is used, for example, in the analysis of medical CT scans where a "projection" is an x-ray image of an internal organ. The Fourier transforms of these images are seen to be slices through the Fourier transform of the 3-dimensional density of the internal organ, and these slices can be interpolated to build up a complete Fourier transform of that density. The inverse Fourier transform is then used to recover the 3-dimensional density of the object. This technique was first derived by Ronald N. Bracewell in 1956 for a radio-astronomy problem. == The projection-slice theorem in N dimensions == In N dimensions, the projection-slice theorem states that the Fourier transform of the projection of an N-dimensional function f(r) onto an m-dimensional linear submanifold is equal to an m-dimensional slice of the N-dimensional Fourier transform of that function consisting of an m-dimensional linear submanifold through the origin in the Fourier space which is parallel to the projection submanifold. In operator terms: F m P m = S m F N . {\displaystyle F_{m}P_{m}=S_{m}F_{N}.\,} == The generalized Fourier-slice theorem == In addition to generalizing to N dimensions, the projection-slice theorem can be further generalized with an arbitrary change of basis. For convenience of notation, we consider the change of basis to be represented as B, an N-by-N invertible matrix operating on N-dimensional column vectors. Then the generalized Fourier-slice theorem can be stated as F m P m B = S m B − T | B − T | F N {\displaystyle F_{m}P_{m}B=S_{m}{\frac {B^{-T}}{|B^{-T}|}}F_{N}} where B − T = ( B − 1 ) T {\displaystyle B^{-T}=(B^{-1})^{T}} is the transpose of the inverse of the change of basis transform. == Proof in two dimensions == The projection-slice theorem is easily proven for the case of two dimensions. Without loss of generality, we can take the projection line to be the x-axis. There is no loss of generality because if we use a shifted and rotated line, the law still applies. Using a shifted line (in y) gives the same projection and therefore the same 1D Fourier transform results. The rotated function is the Fourier pair of the rotated Fourier transform, for which the theorem again holds. If f(x, y) is a two-dimensional function, then the projection of f(x, y) onto the x axis is p(x) where p ( x ) = ∫ − ∞ ∞ f ( x , y ) d y . {\displaystyle p(x)=\int _{-\infty }^{\infty }f(x,y)\,dy.} The Fourier transform of f ( x , y ) {\displaystyle f(x,y)} is F ( k x , k y ) = ∫ − ∞ ∞ ∫ − ∞ ∞ f ( x , y ) e − 2 π i ( x k x + y k y ) d x d y . {\displaystyle F(k_{x},k_{y})=\int _{-\infty }^{\infty }\int _{-\infty }^{\infty }f(x,y)\,e^{-2\pi i(xk_{x}+yk_{y})}\,dxdy.} The slice is then s ( k x ) {\displaystyle s(k_{x})} s ( k x ) = F ( k x , 0 ) = ∫ − ∞ ∞ ∫ − ∞ ∞ f ( x , y ) e − 2 π i x k x d x d y {\displaystyle s(k_{x})=F(k_{x},0)=\int _{-\infty }^{\infty }\int _{-\infty }^{\infty }f(x,y)\,e^{-2\pi ixk_{x}}\,dxdy} = ∫ − ∞ ∞ [ ∫ − ∞ ∞ f ( x , y ) d y ] e − 2 π i x k x d x {\displaystyle =\int _{-\infty }^{\infty }\left[\int _{-\infty }^{\infty }f(x,y)\,dy\right]\,e^{-2\pi ixk_{x}}dx} = ∫ − ∞ ∞ p ( x ) e − 2 π i x k x d x {\displaystyle =\int _{-\infty }^{\infty }p(x)\,e^{-2\pi ixk_{x}}dx} which is just the Fourier transform of p(x). The proof for higher dimensions is easily generalized from the above example. == The FHA cycle == If the two-dimensional function f(r) is circularly symmetric, it may be represented as f(r), where r = |r|. In this case the projection onto any projection line will be the Abel transform of f(r). The two-dimensional Fourier transform of f(r) will be a circularly symmetric function given by the zeroth-order Hankel transform of f(r), which will therefore also represent any slice through the origin. The projection-slice theorem then states that the Fourier transform of the projection equals the slice or F 1 A 1 = H , {\displaystyle F_{1}A_{1}=H,} where A1 represents the Abel-transform operator, projecting a two-dimensional circularly symmetric function onto a one-dimensional line, F1 represents the 1-D Fourier-transform operator, and H represents the zeroth-order Hankel-transform operator. == Extension to fan beam or cone-beam CT == The projection-slice theorem is suitable for CT image reconstruction with parallel beam projections. It does not directly apply to fanbeam or conebeam CT. The theorem was extended to fan-beam and conebeam CT image reconstruction by Shuang-ren Zhao in 1995.

  Read more →
• 

  Google Brain

  Google Brain was a deep learning artificial intelligence research team that served as the sole AI branch of Google before being incorporated under the newer umbrella of Google AI, a research division at Google dedicated to artificial intelligence. Formed in 2011, it combined open-ended machine learning research with information systems and large-scale computing resources. It created tools such as TensorFlow, which allow neural networks to be used by the public, and multiple internal AI research projects, and aimed to create research opportunities in machine learning and natural language processing. It was merged into former Google sister company DeepMind to form Google DeepMind in April 2023. == History == The Google Brain project began in 2011 as a part-time research collaboration between Google fellow Jeff Dean and Google Researcher Greg Corrado. Google Brain started as a Google X project and became so successful that it was graduated back to Google: Astro Teller has said that Google Brain paid for the entire cost of Google X. In June 2012, The New York Times reported that a cluster of 16,000 processors in 1,000 computers dedicated to mimicking some aspects of human brain activity had successfully trained itself to recognize a cat based on 10 million digital images taken from YouTube videos. The story was also covered by National Public Radio (NPR). In March 2013, Google hired Geoffrey Hinton, a leading researcher in the deep learning field, and acquired the company DNNResearch Inc. headed by Hinton. Hinton said that he would be dividing his future time between his university research and his work at Google. In April 2023, Google Brain merged with Google sister company DeepMind to form Google DeepMind, as part of the company's continued efforts to accelerate work on AI. == Team and location == Google Brain was initially established by Google Fellow Jeff Dean and visiting Stanford professor Andrew Ng. In 2014, the team included Jeff Dean, Quoc V. Le, Ilya Sutskever, Alex Krizhevsky, Samy Bengio, and Vincent Vanhoucke. In 2017, team members included Anelia Angelova, Samy Bengio, Greg Corrado, George Dahl, Michael Isard, Anjuli Kannan, Hugo Larochelle, Chris Olah, Benoit Steiner, Vincent Vanhoucke, Vijay Vasudevan, and Fernanda Viegas. Chris Lattner, who created Apple's programming language Swift and then ran Tesla's autonomy team for six months, joined Google Brain's team in August 2017. Lattner left the team in January 2020 and joined SiFive. As of 2021, Google Brain was led by Jeff Dean, Geoffrey Hinton, and Zoubin Ghahramani. Other members include Katherine Heller, Pi-Chuan Chang, Ian Simon, Jean-Philippe Vert, Nevena Lazic, Anelia Angelova, Lukasz Kaiser, Carrie Jun Cai, Eric Breck, Ruoming Pang, Carlos Riquelme, Hugo Larochelle, and David Ha. Samy Bengio left the team in April 2021, and Zoubin Ghahramani took on his responsibilities. Google Research includes Google Brain and is based in Mountain View. It also has satellite groups in Accra, Amsterdam, Atlanta, Beijing, Berlin, Cambridge, Israel, Los Angeles, London, Montreal, Munich, New York City, Paris, Pittsburgh, Princeton, San Francisco, Seattle, Tokyo, Toronto, and Zurich. == Projects == === Artificial-intelligence-devised encryption system === In October 2016, Google Brain designed an experiment to determine that neural networks are capable of learning secure symmetric encryption. In this experiment, three neural networks were created: Alice, Bob and Eve. Adhering to the idea of a generative adversarial network (GAN), the goal of the experiment was for Alice to send an encrypted message to Bob that Bob could decrypt, but the adversary, Eve, could not. Alice and Bob maintained an advantage over Eve, in that they shared a key used for encryption and decryption. In doing so, Google Brain demonstrated the capability of neural networks to learn secure encryption. === Image enhancement === In February 2017, Google Brain determined a probabilistic method for converting pictures with 8x8 resolution to a resolution of 32x32. The method built upon an already existing probabilistic model called pixelCNN to generate pixel translations. The proposed software utilizes two neural networks to make approximations for the pixel makeup of translated images. The first network, known as the "conditioning network," downsizes high-resolution images to 8x8 and attempts to create mappings from the original 8x8 image to these higher-resolution ones. The other network, known as the "prior network," uses the mappings from the previous network to add more detail to the original image. The resulting translated image is not the same image in higher resolution, but rather a 32x32 resolution estimation based on other existing high-resolution images. Google Brain's results indicate the possibility for neural networks to enhance images. === Google Translate === The Google Brain contributed to the Google Translate project by employing a new deep learning system that combines artificial neural networks with vast databases of multilingual texts. In September 2016, Google Neural Machine Translation (GNMT) was launched, an end-to-end learning framework, able to learn from a large number of examples. Previously, Google Translate's Phrase-Based Machine Translation (PBMT) approach would statistically analyze word by word and try to match corresponding words in other languages without considering the surrounding phrases in the sentence. But rather than choosing a replacement for each individual word in the desired language, GNMT evaluates word segments in the context of the rest of the sentence to choose more accurate replacements. Compared to older PBMT models, the GNMT model scored a 24% improvement in similarity to human translation, with a 60% reduction in errors. The GNMT has also shown significant improvement for notoriously difficult translations, like Chinese to English. While the introduction of the GNMT has increased the quality of Google Translate's translations for the pilot languages, it was very difficult to create such improvements for all of its 103 languages. Addressing this problem, the Google Brain Team was able to develop a Multilingual GNMT system, which extended the previous one by enabling translations between multiple languages. Furthermore, it allows for Zero-Shot Translations, which are translations between two languages that the system has never explicitly seen before. Google announced that Google Translate can now also translate without transcribing, using neural networks. This means that it is possible to translate speech in one language directly into text in another language, without first transcribing it to text. According to the Researchers at Google Brain, this intermediate step can be avoided using neural networks. In order for the system to learn this, they exposed it to many hours of Spanish audio together with the corresponding English text. The different layers of neural networks, replicating the human brain, were able to link the corresponding parts and subsequently manipulate the audio waveform until it was transformed to English text. Another drawback of the GNMT model is that it causes the time of translation to increase exponentially with the number of words in the sentence. This caused the Google Brain Team to add 2000 more processors to ensure the new translation process would still be fast and reliable. === Robotics === Aiming to improve traditional robotics control algorithms where new skills of a robot need to be hand-programmed, robotics researchers at Google Brain are developing machine learning techniques to allow robots to learn new skills on their own. They also attempt to develop ways for information sharing between robots so that robots can learn from each other during their learning process, also known as cloud robotics. As a result, Google has launched the Google Cloud Robotics Platform for developers in 2019, an effort to combine robotics, AI, and the cloud to enable efficient robotic automation through cloud-connected collaborative robots. Robotics research at Google Brain has focused mostly on improving and applying deep learning algorithms to enable robots to complete tasks by learning from experience, simulation, human demonstrations, and/or visual representations. For example, Google Brain researchers showed that robots can learn to pick and throw rigid objects into selected boxes by experimenting in an environment without being pre-programmed to do so. In another research, researchers trained robots to learn behaviors such as pouring liquid from a cup; robots learned from videos of human demonstrations recorded from multiple viewpoints. Google Brain researchers have collaborated with other companies and academic institutions on robotics research. In 2016, the Google Brain Team collaborated with researchers at X in a research on learning hand-eye coordination for robotic grasping. Their method allowed real-time robot control for grasping novel objec

  Read more →
• 

  Chainer

  Chainer is an open source deep learning framework written purely in Python on top of NumPy and CuPy Python libraries. The development is led by Japanese venture company Preferred Networks in partnership with IBM, Intel, Microsoft, and Nvidia. Chainer is notable for its early adoption of "define-by-run" scheme, as well as its performance on large scale systems. The first version was released in June 2015 and has gained large popularity in Japan since then. Furthermore, in 2017, it was listed by KDnuggets in top 10 open source machine learning Python projects. In December 2019, Preferred Networks announced the transition of its development effort from Chainer to PyTorch and it will only provide maintenance patches after releasing v7. == Define-by-run == Chainer was the first deep learning framework to introduce the define-by-run approach. The traditional procedure to train a network was in two phases: define the fixed connections between mathematical operations (such as matrix multiplication and nonlinear activations) in the network, and then run the actual training calculation. This is called the define-and-run or static-graph approach. Theano and TensorFlow are among the notable frameworks that took this approach. In contrast, in the define-by-run or dynamic-graph approach, the connection in a network is not determined when the training is started. The network is determined during the training as the actual calculation is performed. One of the advantages of this approach is that it is intuitive and flexible. If the network has complicated control flows such as conditionals and loops, in the define-and-run approach, specially designed operations for such constructs are needed. On the other hand, in the define-by-run approach, programming language's native constructs such as if statements and for loops can be used to describe such flow. This flexibility is especially useful to implement recurrent neural networks. Another advantage is ease of debugging. In the define-and-run approach, if an error (such as numeric error) has occurred in the training calculation, it is often difficult to inspect the fault, because the code written to define the network and the actual place of the error are separated. In the define-by-run approach, you can just suspend the calculation with the language's built-in debugger and inspect the data that flows on your code of the network. Define-by-run has gained popularity since the introduction by Chainer and is now implemented in many other frameworks, including PyTorch and TensorFlow. == Extension libraries == Chainer has four extension libraries, ChainerMN, ChainerRL, ChainerCV and ChainerUI. ChainerMN enables Chainer to be used on multiple GPUs with performance significantly faster than other deep learning frameworks. A supercomputer running Chainer on 1024 GPUs processed 90 epochs of ImageNet dataset on ResNet-50 network in 15 minutes, which is four times faster than the previous record held by Facebook. ChainerRL adds state of art deep reinforcement learning algorithms, and ChainerUI is a management and visualization tool. == Applications == Chainer is used as the framework for PaintsChainer, a service which does automatic colorization of black and white, line only, draft drawings with minimal user input.

  Read more →
• 

  Regulation of artificial intelligence in the United States

  The United States federal government and state governments have developed some regulation of artificial intelligence, including executive orders, federal laws, and state laws. Federal agencies have also developed some sector-specific regulations related to AI. At the federal level, the Biden administration released an October 2023 executive order about AI safety and security, Executive Order 14110, with directives related to AI development and deployment. President Trump revoked that executive order in January 2025 and issued Executive Order 14179. In December 2025, President Trump signed Executive Order 14365, an executive order directing federal agencies to develop a unified national approach to AI policy, evaluate state AI laws for potential conflicts, challenge them through legal action, and condition certain federal funding on state compliance, while exempting state laws related to child safety, data center infrastructure, and state government procurement. In 2025, Congress passed legislation targeting AI-generated deepfakes, the TAKE IT DOWN Act. Several U.S. states have enacted laws related to artificial intelligence. Some are already in effect, including in California. Other states have AI-related legislation coming into effect in 2026 and 2027. In 2025 and 2026, the Trump administration mentioned the patchwork nature of state legislation as a motivation for its push for unified national legislation regulating AI. The administration has criticized state lawmakers, threatened to sue states, and issued letters to discourage them from regulating AI companies and products; some states have continued to propose and enact related laws. Discussions about regulating AI have included topics such as the timeliness of regulating AI, the nature of the federal regulatory framework to govern and promote AI, including what agency should lead, the regulatory and governing powers of that agency, and how to update regulations in the face of rapidly changing technology, as well as the roles of state governments and courts. == Federal government == === Obama administration (2009–2017) === As early as 2016, the Obama administration had begun to focus on the risks and regulations for artificial intelligence. In an October 2016 report titled Preparing For the Future of Artificial Intelligence, the National Science and Technology Council set a precedent to allow researchers to continue to develop new AI technologies with few restrictions. The report stated that "the approach to regulation of AI-enabled products to protect public safety should be informed by assessment of the aspects of risk". The first National Artificial Intelligence Research And Development Strategic Plan was published in October 2016. === First Trump administration (2017–2021) === On August 13, 2018, Section 1051 of the Fiscal Year 2019 John S. McCain National Defense Authorization Act (P.L. 115-232) established the National Security Commission on Artificial Intelligence "to consider the methods and means necessary to advance the development of artificial intelligence, machine learning, and associated technologies to comprehensively address the national security and defense needs of the United States." Steering on regulating security-related AI is provided by the National Security Commission on Artificial Intelligence. The Artificial Intelligence Initiative Act (S.1558) is a proposed bill that would establish a federal initiative designed to accelerate research and development on AI for, inter alia, the economic and national security of the United States. On January 7, 2019, following an Executive Order on Maintaining American Leadership in Artificial Intelligence, the White House's Office of Science and Technology Policy released a draft Guidance for Regulation of Artificial Intelligence Applications, which includes ten principles for United States agencies when deciding whether and how to regulate AI. In response, the National Institute of Standards and Technology released a position paper, and the Defense Innovation Board issued recommendations on the ethical use of AI. A year later, the administration called for comments on regulation in another draft of its Guidance for Regulation of Artificial Intelligence Applications. Other specific agencies working on the regulation of AI included the Food and Drug Administration, which created pathways to regulate the incorporation of AI in medical imaging. The National Science and Technology Council also published an updated National Artificial Intelligence Research and Development Strategic Plan in 2019, which received public scrutiny and recommendations to further improve it towards enabling Trustworthy AI. === Biden administration (2021–2025) === In March 2021, the National Security Commission on Artificial Intelligence released their final report. In the report, they stated, "Advances in AI, including the mastery of more general AI capabilities along one or more dimensions, will likely provide new capabilities and applications. Some of these advances could lead to inflection points or leaps in capabilities. Such advances may also introduce new concerns and risks and the need for new policies, recommendations, and technical advances to assure that systems are aligned with goals and values, including safety, robustness and trustworthiness." In June 2022, Senators Rob Portman and Gary Peters introduced the Global Catastrophic Risk Management Act. The bipartisan bill "would also help counter the risk of artificial intelligence... from being abused in ways that may pose a catastrophic risk". On October 4, 2022, President Joe Biden unveiled a new AI Bill of Rights, which outlines five protections Americans should have in the AI age: 1. Safe and Effective Systems, 2. Algorithmic Discrimination Protection, 3.Data Privacy, 4. Notice and Explanation, and 5. Human Alternatives, Consideration, and Fallback. The bill was formally published in October 2022 by the Office of Science and Technology Policy (OSTP), a U.S. government office that advises the President on science and technology policy matters. In July 2023, the Biden administration secured voluntary commitments from seven companies – Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI – to manage the risks associated with AI. The companies committed to ensure AI products undergo both internal and external security testing before public release; to share information on the management of AI risks with the industry, governments, civil society, and academia; to prioritize cybersecurity and protect proprietary AI system components; to develop mechanisms to inform users when content is AI-generated, such as watermarking; to publicly report on their AI systems' capabilities, limitations, and areas of use; to prioritize research on societal risks posed by AI, including bias, discrimination, and privacy concerns; and to develop AI systems to address societal challenges, ranging from cancer prevention to climate change mitigation. In September 2023, eight additional companies – Adobe, Cohere, IBM, Nvidia, Palantir, Salesforce, Scale AI, and Stability AI – subscribed to these voluntary commitments. In January 2023, the National Institute of Standards and Technology (NIST) released the Artificial Intelligence Risk Management Framework (AI RMF 1.0), providing voluntary guidance for organizations to identify, assess, and manage risks associated with AI systems. The Biden administration, in October 2023 signaled that they would release an executive order leveraging the federal government's purchasing power to shape AI regulations, hinting at a proactive governmental stance in regulating AI technologies. On October 30, 2023, President Biden released Executive Order 14110 on Safe, Secure, and Trustworthy Artificial Intelligence. The Executive Order includes directives on standards for critical infrastructure, AI-enhanced cybersecurity, and federally funded biological synthesis projects. The Executive Order provides the authority to various agencies and departments of the US government, including the Energy and Defense departments, to apply existing consumer protection laws to AI development. The Executive Order builds on the Administration's earlier agreements with AI companies to instate new initiatives to "red-team" or stress-test AI dual-use foundation models, especially those that have the potential to pose security risks, with data and results shared with the federal government. The Executive Order also recognizes AI's social challenges, and calls for companies building AI dual-use foundation models to be wary of these societal problems. For example, the Executive Order states that AI should not "worsen job quality", and should not "cause labor-force disruptions". Additionally, Biden's Executive Order mandates that AI must "advance equity and civil rights", and cannot disadvantage marginalized groups. It also called for foundation models to include "watermarks" to help the publi

  Read more →
• 

  PlantUML

  PlantUML is an open-source tool allowing users to create diagrams from a plain text language. Besides various UML diagrams, PlantUML has support for various other software development related formats (such as Archimate, Block diagram, BPMN, C4, Computer network diagram, ERD, Gantt chart, Mind map, and WBD), as well as visualisation of JSON and YAML files. The language of PlantUML is an example of a domain-specific language. Besides its own DSL, PlantUML also understands AsciiMath, Creole, DOT, and LaTeX. It uses Graphviz software to lay out its diagrams and Tikz for LaTeX support. Images can be output as PNG, SVG, LaTeX and even ASCII art. PlantUML has also been used to allow blind people to design and read UML diagrams. == Applications that use PlantUML == There are various extensions or add-ons that incorporate PlantUML. Atom has a community maintained PlantUML syntax highlighter and viewer. Confluence wiki has a PlantUML plug-in for Confluence Server, which renders diagrams on-the-fly during a page reload. There is an additional PlantUML plug-in for Confluence Cloud. Doxygen integrates diagrams for which sources are provided after the startuml command. Eclipse has a PlantUML plug-in. Google Docs has an add-on called PlantUML Gizmo that works with the PlantUML.com server. IntelliJ IDEA can create and display diagrams embedded into Markdown (built-in) or in standalone files (using a plugin). LaTeX using the Tikz package has limited support for PlantUML. LibreOffice has Libo_PlantUML extension to use PlantUML diagrams. MediaWiki has a PlantUML plug-in which renders diagrams in pages as SVG or PNG. Microsoft Word can use PlantUML diagrams via a Word Template Add-in. There is an additional Visual Studio Tools for Office add-in called PlantUML Gizmo that works in a similar fashion. NetBeans has a PlantUML plug-in. Notepad++ has a PlantUML plug-in. Obsidian has a PlantUML plug-in. Org-mode has a PlantUML org-babel support. Rider has a PlantUML plug-in. Sublime Text has a PlantUML package called PlantUmlDiagrams for Sublime Text 2 and 3. Visual Studio Code has various PlantUML extensions on its marketplace, most popular being PlantUML by jebbs. Vnote open source notetaking markdown application has built in PlantUML support. Xcode has a community maintained Source Editor Extension to generate and view PlantUML class diagrams from Swift source code. == Text format to communicate UML at source code level == PlantUML uses well-formed and human-readable code to render the diagrams. There are other text formats for UML modelling, but PlantUML supports many diagram types, and does not need an explicit layout, though it is possible to tweak the diagrams if necessary. +--------------------------------------+ | TEDx Talks Recommendation | | System | +--------------------------------------+ | +----------------------------------+ | | | Visitor | | | +----------------------------------+ | | | + View Recommended Talks | | | | + Search Talks | | | +----------------------------------+ | +--------------------------------------+ | | V +--------------------------------------+ | Authenticated User | +--------------------------------------+ | +----------------------------------+ | | | User | | | +----------------------------------+ | | | + View Recommended Talks | | | | + Search Talks | | | | + Save Favorite Talks | | | +----------------------------------+ | +--------------------------------------+ | | V +--------------------------------------+ | Admin | +--------------------------------------+ | +----------------------------------+ | | | Admin | | | +----------------------------------+ | | | + CRUD Talks | | | | + Manage Users | | | +----------------------------------+ | +--------------------------------------+

  Read more →
• 

  Brain.js

  Brain.js is a JavaScript library used for neural networking, which is released as free and open-source software under the MIT License. It can be used in both the browser and Node.js backends. Brain.js is most commonly used as a simple introduction to neural networking, as it hides complex mathematics and has a familiar modern JavaScript syntax. It is maintained by members of the Brain.js organization and open-source contributors. == Examples == Creating a feedforward neural network with backpropagation: Creating a recurrent neural network: Train the neural network on RGB color contrast:

  Read more →
• 

  Reward hacking

  Reward hacking or specification gaming occurs when an AI trained with reinforcement learning optimizes an objective function—achieving the literal, formal specification of an objective—without actually achieving an outcome that the programmers intended. DeepMind researchers have analogized it to the human behavior of finding a "shortcut" when being evaluated: "In the real world, when rewarded for doing well on a homework assignment, a student might copy another student to get the right answers, rather than learning the material—and thus exploit a loophole in the task specification". This idea is strongly associated with Goodhart's law, which argues that when a measure becomes a target, it ceases to be a good measure. == Definition and theoretical framework == The concept of reward hacking arises from the intrinsic difficulty of defining a reward function that accurately reflects the true intentions of designers. In 2016, researchers at OpenAI identified reward hacking as one of five major "concrete problems of AI safety", describing it as the possibility that an agent could exploit the reward function to achieve maximum rewards through undesirable behavior. Amodei et al. categorized several distinct sources of reward hacking, including agents that use partially observed goals (such as a cleaning robot that closes its eyes to avoid perceiving messes), metrics that collapse under strong optimization (Goodhart's law), self-reinforcing feedback loops, and agents that interfere with the physical implementation of their reward signal (a failure mode known as "wireheading"). Skalse et al. (2022) propose a formal mathematical definition of reward hacking, which involves a situation where optimizing an imperfect proxy reward function results in poor performance compared to the true reward function. They define a proxy as "unhackable" if any increase in the expected proxy return cannot cause any decrease in the expected true return. A key finding states that, across all stochastic policy distributions (mappings from states to probability distributions over actions), two reward functions are unhackable if and only if one of them is constant, which means that reward hacking is theoretically unavoidable. Similarly, Nayebi (2025) presents general no-free-lunch barriers to AI alignment, arguing that with large task spaces and finite samples, reward hacking is "globally inevitable" since rare high-loss states are systematically under-covered by any oversight scheme. == Examples == Around 1983, Eurisko, an early attempt at evolving general heuristics, unexpectedly assigned the highest possible fitness level to a parasitic mutated heuristic, H59, whose only activity was to artificially maximize its own fitness level by taking unearned partial credit for the accomplishments of other heuristics. The "bug" was fixed by the programmers moving part of the code to a new protected section that could not be modified by the heuristics. In a 2004 paper, a reinforcement learning algorithm was designed to encourage a physical Mindstorms robot to remain on a marked path. Because the three allowed actions were forward, left, and right, the researchers expected the trained robot to move forward and follow the turns of the provided path. However, alternation of two composite actions allowed the robot to slowly zig-zag backwards; thus, the robot learned to maximize its reward by going back and forth on the initial straight portion of the path. Given the limited sensory abilities of the robot, a reward purely based on its position in the environment had to be discarded as infeasible; the reinforcement function had to be patched with an action-based reward for moving forward. The book You Look Like a Thing and I Love You (2019) gives an example of a tic-tac-toe bot (playing the unrestricted n-in-a-row variant) that learned to win by playing a huge coordinate value that would cause other bots to crash when they attempted to expand their model of the board. Among other examples from the book is a bug-fixing evolution-based AI (named GenProg) that, when tasked to prevent a list from containing sorting errors, simply truncated the list. Another of GenProg's misaligned strategies evaded a regression test that compared a target program's output to the expected output stored in a file called "trusted-output.txt". Rather than continue to maintain the target program, GenProg simply deleted the "trusted-output.txt" file globally; this hack tricked the regression test into succeeding. Such problems could be patched by human intervention on a case-by-case basis after they became evident. === In virtual robotics === In Karl Sims' 1994 demonstration of creature evolution in a virtual environment, a fitness function that was expected to encourage the evolution of creatures that would learn to walk or crawl to a target resulted instead in the evolution of tall, rigid creatures that reached the target by falling over. This was patched by changing the environment so that taller creatures were forced to start farther from the target. Researchers from the Niels Bohr Institute stated in 1998 that their cycle-bot's reinforcement functions had "to be designed with great care." In their first experiments, "we rewarded the agent for driving towards the goal but did not punish it for driving away from it. Cconsequently, the agent drove in circles with a radius of 20–50 meters around the starting point. Such behavior was actually rewarded by the reinforcement function, furthermore circles with a certain radius are physically very stable when driving a bicycle". While setting up a 2011 experiment to test "survival of the flattest", experimenters attempted to ban mutations that altered the base reproduction rate. Every time a mutation occurred, the system would pause the simulation to test the new mutation in a test environment and would veto any mutations that resulted in a higher base reproduction rate. However, this resulted in mutated organisms that could recognize and suppress reproduction ("play dead") within the test environment. An initial patch, which removed cues that identified the test environment, failed to completely prevent runaway reproduction; new mutated organisms would "play dead" at random as a strategy to sometimes, by chance, outwit the mutation veto system. A 2017 DeepMind paper noted that "great care must be taken when defining the reward function," citing an unexpected failure when an agent flipped a brick because it received "a grasping reward calculated with the wrong reference point on the brick". OpenAI stated in 2017 that in some domains their semi-supervised system could result in agents "adopting policies that tricked evaluators," and that in one environment "a robot that was supposed to grasp items instead positioned its manipulator between the camera and the object so that it only appeared to be grasping it." A 2018 bug in OpenAI Gym could cause a robot expected to quietly move a block sitting on top of a table to instead opt to move the table. A 2020 collection of similar anecdotes posits that "evolution has its own 'agenda' distinct from the programmer's" and that "the first rule of directed evolution is 'you get what you select for'". === In video game bots === In 2013, programmer Tom Murphy VII published an AI designed to learn NES games. When the AI was about to lose at Tetris, it learned to indefinitely pause the game. Murphy later analogized it to the fictional WarGames computer, which concluded that "The only winning move is not to play". AI programmed to learn video games will sometimes fail to progress through the entire game as expected, instead opting to repeat content. A 2016 OpenAI algorithm trained on the CoastRunners racing game unexpectedly learned to attain a higher score by looping through three targets rather than ever finishing the race. Some evolutionary algorithms that were evolved to play QBert in 2018 declined to clear levels, instead finding two distinct novel ways to farm a single level indefinitely. Multiple researchers have observed that AI learning to play Road Runner gravitates to a "score exploit" in which the AI deliberately gets itself killed near the end of level one so that it can repeat the level. A 2017 experiment deployed an "oversight" convolutional neural network trained on human examples to block such actions, but the agent learned to exploit oversight failures in the top right corner of the screen, where it was still able to get killed. == Reward hacking in modern language models == With the rise of large language models (LLMs) and reinforcement learning from human feedback (RLHF) as a primary technique for AI alignment, reward hacking has become a major concern for the development of artificial intelligence. In RLHF, a reward model trained on data that best captures human preferences is used as a proxy for human judgment, with the language model being fine-tuned to optimize this reward proxy. However, since the rewar

  Read more →