AI Face Korean

AI Face Korean — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

Time-inhomogeneous hidden Bernoulli model

Time-inhomogeneous hidden Bernoulli model (TI-HBM) is an alternative to hidden Markov model (HMM) for automatic speech recognition. Contrary to HMM, the state transition process in TI-HBM is not a Markov-dependent process, rather it is a generalized Bernoulli (an independent) process. This difference leads to elimination of dynamic programming at state-level in TI-HBM decoding process. Thus, the computational complexity of TI-HBM for probability evaluation and state estimation is O ( N L ) {\displaystyle O(NL)} (instead of O ( N 2 L ) {\displaystyle O(N^{2}L)} in the HMM case, where N {\displaystyle N} and L {\displaystyle L} are number of states and observation sequence length respectively). The TI-HBM is able to model acoustic-unit duration (e.g. phone/word duration) by using a built-in parameter named survival probability. The TI-HBM is simpler and faster than HMM in a phoneme recognition task, but its performance is comparable to HMM. For details, see [1] or [2].
Read more →
Run-time algorithm specialization

In computer science, run-time algorithm specialization is a methodology for creating efficient algorithms for costly computation tasks of certain kinds. The methodology originates in the field of automated theorem proving and, more specifically, in the Vampire theorem prover project. The idea is inspired by the use of partial evaluation in optimising program translation. Many core operations in theorem provers exhibit the following pattern. Suppose that we need to execute some algorithm a l g ( A , B ) {\displaystyle {\mathit {alg}}(A,B)} in a situation where a value of A {\displaystyle A} is fixed for potentially many different values of B {\displaystyle B} . In order to do this efficiently, we can try to find a specialization of a l g {\displaystyle {\mathit {alg}}} for every fixed A {\displaystyle A} , i.e., such an algorithm a l g A {\displaystyle {\mathit {alg}}_{A}} , that executing a l g A ( B ) {\displaystyle {\mathit {alg}}_{A}(B)} is equivalent to executing a l g ( A , B ) {\displaystyle {\mathit {alg}}(A,B)} . The specialized algorithm may be more efficient than the generic one, since it can exploit some particular properties of the fixed value A {\displaystyle A} . Typically, a l g A ( B ) {\displaystyle {\mathit {alg}}_{A}(B)} can avoid some operations that a l g ( A , B ) {\displaystyle {\mathit {alg}}(A,B)} would have to perform, if they are known to be redundant for this particular parameter A {\displaystyle A} . In particular, we can often identify some tests that are true or false for A {\displaystyle A} , unroll loops and recursion, etc. == Difference from partial evaluation == The key difference between run-time specialization and partial evaluation is that the values of A {\displaystyle A} on which a l g {\displaystyle {\mathit {alg}}} is specialised are not known statically, so the specialization takes place at run-time. There is also an important technical difference. Partial evaluation is applied to algorithms explicitly represented as codes in some programming language. At run-time, we do not need any concrete representation of a l g {\displaystyle {\mathit {alg}}} . We only have to imagine a l g {\displaystyle {\mathit {alg}}} when we program the specialization procedure. All we need is a concrete representation of the specialized version a l g A {\displaystyle {\mathit {alg}}_{A}} . This also means that we cannot use any universal methods for specializing algorithms, which is usually the case with partial evaluation. Instead, we have to program a specialization procedure for every particular algorithm a l g {\displaystyle {\mathit {alg}}} . An important advantage of doing so is that we can use some powerful ad hoc tricks exploiting peculiarities of a l g {\displaystyle {\mathit {alg}}} and the representation of A {\displaystyle A} and B {\displaystyle B} , which are beyond the reach of any universal specialization methods. == Specialization with compilation == The specialized algorithm has to be represented in a form that can be interpreted. In many situations, usually when a l g A ( B ) {\displaystyle {\mathit {alg}}_{A}(B)} is to be computed on many values of B {\displaystyle B} in a row, a l g A {\displaystyle {\mathit {alg}}_{A}} can be written as machine code instructions for a special abstract machine, and it is typically said that A {\displaystyle A} is compiled. The code itself can then be additionally optimized by answer-preserving transformations that rely only on the semantics of instructions of the abstract machine. The instructions of the abstract machine can usually be represented as records. One field of such a record, an instruction identifier (or instruction tag), would identify the instruction type, e.g. an integer field may be used, with particular integer values corresponding to particular instructions. Other fields may be used for storing additional parameters of the instruction, e.g. a pointer field may point to another instruction representing a label, if the semantics of the instruction require a jump. All instructions of the code can be stored in a traversable data structure such as an array, linked list, or tree. Interpretation (or execution) proceeds by fetching instructions in some order, identifying their type, and executing the actions associated with said type. In many programming languages, such as C and C++, a simple switch statement may be used to associate actions with different instruction identifiers. Modern compilers usually compile a switch statement with constant (e.g. integer) labels from a narrow range by storing the address of the statement corresponding to a value i {\displaystyle i} in the i {\displaystyle i} -th cell of a special array, as a means of efficient optimisation. This can be exploited by taking values for instruction identifiers from a small interval of values. == Data-and-algorithm specialization == There are situations when many instances of A {\displaystyle A} are intended for long-term storage and the calls of a l g ( A , B ) {\displaystyle {\mathit {alg}}(A,B)} occur with different B {\displaystyle B} in an unpredictable order. For example, we may have to check a l g ( A 1 , B 1 ) {\displaystyle {\mathit {alg}}(A_{1},B_{1})} first, then a l g ( A 2 , B 2 ) {\displaystyle {\mathit {alg}}(A_{2},B_{2})} , then a l g ( A 1 , B 3 ) {\displaystyle {\mathit {alg}}(A_{1},B_{3})} , and so on. In such circumstances, full-scale specialization with compilation may not be suitable due to excessive memory usage. However, we can sometimes find a compact specialized representation A ′ {\displaystyle A^{\prime }} for every A {\displaystyle A} , that can be stored with, or instead of, A {\displaystyle A} . We also define a variant a l g ′ {\displaystyle {\mathit {alg}}^{\prime }} that works on this representation and any call to a l g ( A , B ) {\displaystyle {\mathit {alg}}(A,B)} is replaced by a l g ′ ( A ′ , B ) {\displaystyle {\mathit {alg}}^{\prime }(A^{\prime },B)} , intended to do the same job faster.
Read more →
UI data binding

UI data binding is a software design pattern to simplify development of GUI applications. UI data binding binds UI elements to an application domain model. Most frameworks employ the Observer pattern as the underlying binding mechanism. To work efficiently, UI data binding has to address input validation and data type mapping. A bound control is a widget whose value is tied or bound to a field in a recordset (e.g., a column in a row of a table). Changes made to data within the control are automatically saved to the database when the control's exit event triggers. == Example == == Data binding frameworks and tools == === Delphi === DSharp third-party data binding tool OpenWire Visual Live Binding - third-party visual data binding tool === Java === JFace Data Binding JavaFX Property === .NET === Windows Forms data binding overview WPF data binding overview Avalonia Unity 3D data binding framework (available in modifications for NGUI, iGUI and EZGUI libraries) === JavaScript === Angular AngularJS Backbone.js Ember.js Datum.js knockout.js Meteor, via its Blaze live update engine OpenUI5 React Vue.js
Read more →
Wearable technology

Wearable technology is a category of small electronic and mobile devices with wireless communications capability designed to be worn on the human body and are incorporated into gadgets, accessories, or clothes. Common types of wearable technology include smartwatches, fitness trackers, and smartglasses. Wearable electronic devices are often close to or on the surface of the skin, where they detect, analyze, and transmit information such as vital signs, and/or ambient data and which allow in some cases immediate biofeedback to the wearer. Wearable devices collect vast amounts of data from users making use of different behavioral and physiological sensors, which monitor their health status and activity levels. Wrist-worn devices include smartwatches with a touchscreen display, while wristbands are mainly used for fitness tracking but do not contain a touchscreen display. Wearable devices such as activity trackers are an example of the Internet of things, since "things" such as electronics, software, sensors, and connectivity are effectors that enable objects to exchange data (including data quality) through the internet with a manufacturer, operator, and/or other connected devices, without requiring human intervention. Wearable technology offers a wide range of possible uses, from communication and entertainment to improving health and fitness, however, there are worries about privacy and security because wearable devices have the ability to collect personal data. Wearable technology has a variety of use cases which is growing as the technology is developed and the market expands. It can be used to encourage individuals to be more active and improve their lifestyle choices. Healthy behavior is encouraged by tracking activity levels and providing useful feedback to enable goal setting. This can be shared with interested stakeholders such as healthcare providers. Wearables are popular in consumer electronics, most commonly in the form factors of smartwatches, smart rings, and implants. Apart from commercial uses, wearable technology is being incorporated into navigation systems, advanced textiles (e-textiles), and healthcare. As wearable technology is being proposed for use in critical applications, like other technology, it is vetted for its reliability and security properties. == History == In the 1500s, German inventor Peter Henlein (1485–1542) created small watches that were worn as necklaces. A century later, pocket watches grew in popularity as waistcoats became fashionable for men. Wristwatches were created in the late 1600s but were worn mostly by women as bracelets. Pedometers were developed around the same time as pocket watches. The concept of a pedometer was described by Leonardo da Vinci around 1500, and the Germanic National Museum in Nuremberg has a pedometer in its collection from 1590. In the late 1800s, the first wearable hearing aids were introduced. In 1904, aviator Alberto Santos-Dumont pioneered the modern use of the wristwatch. In 1949, American biophysicist Norman Holter invented the very first health monitoring device. His invention, the Holter monitor, was groundbreaking as one of the first wearable devices capable of tracking vital health data outside of a clinical setting. In the 1970s, calculator watches became available, reaching the peak of their popularity in the 1980s. From the early 2000s, wearable cameras were being used as part of a growing sousveillance movement. Expectations, operations, usage and concerns about wearable technology was floated on the first International Conference on Wearable Computing. In 2008, Ilya Fridman incorporated a hidden Bluetooth microphone into a pair of earrings. Big tech companies such as Apple, Samsung, and Fitbit have expanded on this idea by interfacing with smartphones and personal computer software to collect a wide variety of data. Wearable devices include dedicated health monitors, fitness bands, and smartwatches. In 2010, Fitbit released its first step counter. Wearable technology which tracks information such as walking and heart rate is part of the quantified self movement. In 2013, McLear, also known as NFC Ring, released a "smart ring". The smart ring could make bitcoin payments, unlock other devices, and transfer personally identifying information, and also had other features. In 2013, one of the first widely available smartwatches was the Samsung Galaxy Gear. Apple followed in 2015 with the Apple Watch. === Prototypes === From 1991 to 1997, Rosalind Picard and her students, Steve Mann and Jennifer Healey, at the MIT Media Lab designed, built, and demonstrated data collection and decision making from "Smart Clothes" that monitored continuous physiological data from the wearer. These "smart clothes", "smart underwear", "smart shoes", and smart jewellery collected data that related to affective state and contained or controlled physiological sensors and environmental sensors like cameras and other devices. At the same time, also at the MIT Media Lab, Thad Starner and Alex "Sandy" Pentland develop augmented reality. In 1997, their smartglass prototype is featured on 60 Minutes and enables rapid web search and instant messaging. Though the prototype's glasses are nearly as streamlined as modern smartglasses, the processor was a computer worn in a backpack – the most lightweight solution available at the time. In 2009, Sony Ericsson teamed up with the London College of Fashion for a contest to design digital clothing. The winner was a cocktail dress with Bluetooth technology making it light up when a call is received. Zach "Hoeken" Smith of MakerBot fame made keyboard pants during a "Fashion Hacking" workshop at a New York City creative collective. The Tyndall National Institute in Ireland developed a "remote non-intrusive patient monitoring" platform which was used to evaluate the quality of the data generated by the patient sensors and how the end users may adopt to the technology. More recently, London-based fashion company CuteCircuit created costumes for singer Katy Perry featuring LED lighting so that the outfits would change color both during stage shows and appearances on the red carpet such as the dress Katy Perry wore in 2010 at the MET Gala in NYC. In 2012, CuteCircuit created the world's first dress to feature Tweets, as worn by singer Nicole Scherzinger. In 2010, McLear, also known as NFC Ring, developed prototypes of its "smart ring" devices, before a Kickstarter fundraising in 2013. In 2014, graduate students from the Tisch School of Arts in New York designed a hoodie that sent pre-programmed text messages triggered by gesture movements. Around the same time, prototypes for digital eyewear with heads up display (HUD) began to appear. The US military employs headgear with displays for soldiers using a technology called holographic optics. In 2010, Google started developing prototypes of its optical head-mounted display Google Glass, which went into customer beta in March 2013. == Usage == In the consumer space, sales of smart wristbands (aka activity trackers such as the Jawbone UP and Fitbit Flex) started accelerating in 2013. One in five American adults have a wearable device, according to the 2014 PriceWaterhouseCoopers Wearable Future Report. As of 2009, decreasing cost of processing power and other components was facilitating widespread adoption and availability. In professional sports, wearable technology has applications in monitoring and real-time feedback for athletes. Examples of wearable technology in sport include accelerometers, pedometers, and GPS's which can be used to measure an athlete's energy expenditure and movement pattern. In cybersecurity and financial technology, secure wearable devices have captured part of the physical security key market. McLear, also known as NFC Ring, and VivoKey developed products with one-time pass secure access control. In health informatics, wearable devices have enabled better capturing of human health statistics for data driven analysis. This has facilitated data-driven machine learning algorithms to analyse the health condition of users. In business, wearable technology helps managers easily supervise employees by knowing their locations and what they are currently doing. Employees working in a warehouse also have increased safety when working around chemicals or lifting something. Smart helmets are employee safety wearables that have vibration sensors that can alert employees of possible danger in their environment. == Wearable technology and health == Wearable technology is often used to monitor a user's health. Given that such a device is in close contact with the user, it can easily collect data. It started as soon as 1980 where first wireless ECG was invented. In the last decades, there has been substantial growth in research of e.g. textile-based, tattoo, patch, and contact lenses as well as circulation of a notion of "quantified self", transhumanism-related ideas, and growth of life ex
Read more →
Computational heuristic intelligence

Computational heuristic intelligence (CHI) refers to specialized programming techniques in computational intelligence (also called artificial intelligence, or AI). These techniques have the express goal of avoiding complexity issues, also called NP-hard problems, by using human-like techniques. They are best summarized as the use of exemplar-based methods (heuristics), rather than rule-based methods (algorithms). Hence the term is distinct from the more conventional computational algorithmic intelligence, or symbolic AI. An example of a CHI technique is the encoding specificity principle of Tulving and Thompson. In general, CHI principles are problem solving techniques used by people, rather than programmed into machines. It is by drawing attention to this key distinction that the use of this term is justified in a field already replete with confusing neologisms. Note that the legal systems of all modern human societies employ both heuristics (generalisations of cases) from individual trial records as well as legislated statutes (rules) as regulatory guides. Another recent approach to the avoidance of complexity issues is to employ feedback control rather than feedforward modeling as a problem-solving paradigm. This approach has been called computational cybernetics, because (a) the term 'computational' is associated with conventional computer programming techniques which represent a strategic, compiled, or feedforward model of the problem, and (b) the term 'cybernetic' is associated with conventional system operation techniques which represent a tactical, interpreted, or feedback model of the problem. Of course, real programs and real problems both contain both feedforward and feedback components. A real example which illustrates this point is that of human cognition, which clearly involves both perceptual (bottom-up, feedback, sensor-oriented) and conceptual (top-down, feedforward, motor-oriented) information flows and hierarchies. The AI engineer must choose between mathematical and cybernetic problem solution and machine design paradigms. This is not a coding (program language) issue, but relates to understanding the relationship between the declarative and procedural programming paradigms. The vast majority of STEM professionals never get the opportunity to design or implement pure cybernetic solutions. When pushed, most responders will dismiss the importance of any difference by saying that all code can be reduced to a mathematical model anyway. Unfortunately, not only is this belief false, it fails most spectacularly in many AI scenarios. Mathematical models are not time agnostic, but by their very nature are pre-computed, i.e. feedforward. Dyer [2012] and Feldman [2004] have independently investigated the simplest of all somatic governance paradigms, namely control of a simple jointed limb by a single flexor muscle. They found that it is impossible to determine forces from limb positions- therefore, the problem cannot have a pre-computed (feedforward) mathematical solution. Instead, a top-down command bias signal changes the threshold feedback level in the sensorimotor loop, e.g. the loop formed by the afferent and efferent nerves, thus changing the so-called ‘equilibrium point’ of the flexor muscle/ elbow joint system. An overview of the arrangement reveals that global postures and limb position are commanded in feedforward terms, using global displacements (common coding), with the forces needed being computed locally by feedback loops. This method of sensorimotor unit governance, which is based upon what Anatol Feldman calls the ‘equilibrium Point’ theory, is formally equivalent to a servomechanism such as a car's ‘cruise control’.
Read more →
TurboQuant

TurboQuant is an online vector quantization algorithm for compressing high-dimensional Euclidean vectors while preserving their geometric structure. It was proposed in 2025 by Amir Zandieh, Majid Daliri, Majid Hadian, and Vahab Mirrokni in the paper TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate. The paper lists Zandieh and Mirrokni as affiliated with Google Research, Daliri with New York University, and Hadian with Google DeepMind. The method was developed for applications including large language model (LLM) inference, key–value (KV) cache compression, vector databases, and nearest neighbor search. TurboQuant consists of two related algorithms: TurboQuantmse, which is optimized for mean squared error (MSE), and TurboQuantprod, which is optimized for unbiased inner product estimation. The algorithm uses a random rotation of input vectors, applies scalar quantizers to the rotated coordinates, and, for inner-product estimation, applies a one-bit Quantized Johnson–Lindenstrauss (QJL) transform to the residual error. == Background == Vector quantization is a compression method that maps high-dimensional vectors to a finite set of codewords. The problem has roots in Shannon's source coding theory and rate–distortion theory. In machine learning and information retrieval, vector quantization is used to reduce the memory required to store embeddings, activation vectors, and other numerical representations. In Transformer-based large language models, the KV cache stores key and value vectors from previous tokens during autoregressive decoding. The size of this cache grows with context length, the number of attention heads, and the number of concurrent requests, making it a major memory bottleneck in LLM serving. Similar compression problems appear in vector search, where large collections of embedding vectors must be stored and searched efficiently. Earlier approaches to vector quantization include product quantization, scalar quantization, and data-dependent k-means codebook construction. The TurboQuant paper argues that many existing methods either require offline preprocessing and calibration or suffer from suboptimal distortion guarantees in online settings. == Algorithm == === TurboQuantmse === TurboQuantmse is the version of the algorithm optimized for mean-squared error. For a unit vector x ∈ S d − 1 {\displaystyle x\in S^{d-1}} , the algorithm first applies a random rotation matrix Π ∈ R d × d {\displaystyle \Pi \in \mathbb {R} ^{d\times d}} and sets z = Π x {\displaystyle z=\Pi x} . Each coordinate of the rotated vector follows a shifted and scaled beta distribution, which converges to a normal distribution in high dimensions. In high dimensions, distinct coordinates also become nearly independent, allowing the algorithm to apply scalar quantizers independently to each coordinate. The scalar quantizer is constructed by solving a one-dimensional continuous k-means or Lloyd–Max quantization problem. If the centroids are c 1 , c 2 , … , c 2 b {\displaystyle c_{1},c_{2},\ldots ,c_{2^{b}}} , the quantization step stores, for each coordinate, i d x j = ⁡ a r g m i n k ∈ [ 2 b ] | z j − c k | . {\displaystyle \mathrm {idx} _{j}=\operatorname {} {arg\,min}_{k\in [2^{b}]}|z_{j}-c_{k}|.} During dequantization, the stored index for each coordinate is replaced by the corresponding centroid, giving a reconstructed rotated vector z ~ {\displaystyle {\tilde {z}}} . The algorithm then rotates back: x ~ = Π ⊤ z ~ . {\displaystyle {\tilde {x}}=\Pi ^{\top }{\tilde {z}}.} The paper gives the following bound for TurboQuantmse: D m s e ≤ 3 π 2 ⋅ 1 4 b . {\displaystyle D_{\mathrm {mse} }\leq {\frac {\sqrt {3\pi }}{2}}\cdot {\frac {1}{4^{b}}}.} It also reports finer-grained MSE values of approximately 0.36, 0.117, 0.03, and 0.009 for bit-widths b = 1 , 2 , 3 , 4 {\displaystyle b=1,2,3,4} , respectively. === TurboQuantprod === TurboQuantprod is optimized for unbiased inner-product estimation. The authors note that an MSE-optimized quantizer may introduce bias when used to estimate inner products. To address this, TurboQuantprod first applies TurboQuantmse with bit-width b − 1 {\displaystyle b-1} , then applies a one-bit Quantized Johnson–Lindenstrauss transform to the remaining residual vector. Let r = x − Q m s e − 1 ( Q m s e ( x ) ) {\displaystyle r=x-Q_{\mathrm {mse} }^{-1}(Q_{\mathrm {mse} }(x))} be the residual after MSE quantization, and let γ = ‖ r ‖ 2 {\displaystyle \gamma =\|r\|_{2}} . The QJL step stores a sign vector for the residual. For γ ≠ 0 {\displaystyle \gamma \neq 0} , this can be written using the normalized residual u = r / γ {\displaystyle u=r/\gamma } : q j l = sign ⁡ ( S u ) , {\displaystyle qjl=\operatorname {sign} (Su),} where S ∈ R d × d {\displaystyle S\in \mathbb {R} ^{d\times d}} is a random projection matrix. Since the sign function is invariant under positive rescaling, this is equivalent to sign ⁡ ( S r ) {\displaystyle \operatorname {sign} (Sr)} when r ≠ 0 {\displaystyle r\neq 0} . If γ = 0 {\displaystyle \gamma =0} , the residual correction is zero. TurboQuantprod stores the MSE quantization, the QJL sign vector, and the residual norm: Q p r o d ( x ) = [ Q m s e ( x ) , q j l , γ ] . {\displaystyle Q_{\mathrm {prod} }(x)=\left[Q_{\mathrm {mse} }(x),qjl,\gamma \right].} The dequantized vector is reconstructed as x ~ = x ~ m s e + π / 2 d γ S ⊤ q j l . {\displaystyle {\tilde {x}}={\tilde {x}}_{\mathrm {mse} }+{\frac {\sqrt {\pi /2}}{d}}\,\gamma S^{\top }qjl.} The paper proves that TurboQuantprod is unbiased for inner-product estimation: E x ~ [ ⟨ y , x ~ ⟩ ] = ⟨ y , x ⟩ . {\displaystyle \mathbb {E} _{\tilde {x}}\left[\langle y,{\tilde {x}}\rangle \right]=\langle y,x\rangle .} It also gives the distortion bound D p r o d ≤ 3 π 2 ⋅ ‖ y ‖ 2 2 d ⋅ 1 4 b . {\displaystyle D_{\mathrm {prod} }\leq {\frac {\sqrt {3\pi }}{2}}\cdot {\frac {\|y\|_{2}^{2}}{d}}\cdot {\frac {1}{4^{b}}}.} == Performance and applications == The TurboQuant paper reports that the algorithm achieves near-optimal distortion rates within a small constant factor of information-theoretic lower bounds. The authors report that, for KV cache quantization, TurboQuant achieved quality neutrality at 3.5 bits per channel and marginal degradation at 2.5 bits per channel. In long-context LLM experiments using Llama 3.1 8B Instruct, the paper evaluated the method on a "needle-in-a-haystack" retrieval task with document lengths from 4,000 to 104,000 tokens. It reported that TurboQuant matched the uncompressed full-precision baseline while using more than 4× compression, and compared the method against PolarQuant, SnapKV, PyramidKV, and KIVI. Google Research stated that TurboQuant was evaluated on long-context benchmarks including LongBench, Needle in a Haystack, ZeroSCROLLS, RULER, and L-Eval using open-source models including Gemma and Mistral. According to a report in Tom's Hardware, Google described the method as reducing KV-cache memory by at least six times and achieving up to an eightfold improvement in attention-logit computation on Nvidia H100 GPUs compared with unquantized 32-bit keys. TurboQuant has also been applied to nearest-neighbor vector search. The original paper reports experiments on DBpedia entity embeddings and GloVe embeddings, comparing TurboQuant with product quantization and other vector-search quantization baselines. == Relationship to other methods == TurboQuant is related to several methods for efficient large language model inference and high-dimensional search: Product quantization – a vector quantization technique widely used for approximate nearest-neighbor search Quantization (machine learning) – reducing the numerical precision of weights, activations, or cached tensors in machine learning models PagedAttention – a memory-management algorithm for LLM serving that reduces fragmentation in the KV cache Johnson–Lindenstrauss lemma – a result in high-dimensional geometry used in random projection methods Lloyd's algorithm – an algorithm for scalar and vector quantization, including k-means-style codebook construction Unlike PagedAttention, which focuses on memory allocation and cache layout, TurboQuant reduces the numerical storage cost of the vectors themselves. Unlike many product-quantization methods, TurboQuant is designed to be data-oblivious and online, avoiding dataset-specific codebook training. == Limitations == The strongest performance claims for TurboQuant come from the original paper and Google Research's own publication. Coverage in technology media has noted that the broader impact of the method will depend on real-world implementation details, workloads, and hardware architectures.
Read more →
OpenSMILE

openSMILE is source-available software for automatic extraction of features from audio signals and for classification of speech and music signals. "SMILE" stands for "Speech & Music Interpretation by Large-space Extraction". The software is mainly applied in the area of automatic emotion recognition and is widely used in the affective computing research community. The openSMILE project exists since 2008 and is maintained by the German company audEERING GmbH since 2013. openSMILE is provided free of charge for research purposes and personal use under a source-available license. For commercial use of the tool, the company audEERING offers custom license options. == Application Areas == openSMILE is used for academic research as well as for commercial applications in order to automatically analyze speech and music signals in real-time. In contrast to automatic speech recognition which extracts the spoken content out of a speech signal, openSMILE is capable of recognizing the characteristics of a given speech or music segment. Examples for such characteristics encoded in human speech are a speaker's emotion, age, gender, and personality, as well as speaker states like depression, intoxication, or vocal pathological disorders. The software further includes music classification technology for automatic music mood detection and recognition of chorus segments, key, chords, tempo, meter, dance-style, and genre. The openSMILE toolkit serves as benchmark in manifold research competitions such as Interspeech ComParE, AVEC, MediaEval, and EmotiW. == History == The openSMILE project was started in 2008 by Florian Eyben, Martin Wöllmer, and Björn Schuller at the Technical University of Munich within the European Union research project SEMAINE. The goal of the SEMAINE project was to develop a virtual agent with emotional and social intelligence. In this system, openSMILE was applied for real-time analysis of speech and emotion. The final SEMAINE software release is based on openSMILE version 1.0.1. In 2009, the emotion recognition toolkit (openEAR) was published based on openSMILE. "EAR" stands for "Emotion and Affect Recognition". In 2010, openSMILE version 1.0.1 was published and was introduced and awarded at the ACM Multimedia Open-Source Software Challenge. Between 2011 and 2013, the technology of openSMILE was extended and improved by Florian Eyben and Felix Weninger in the context of their doctoral thesis at the Technical University of Munich. The software was also applied for the project ASC-Inclusion, which was funded by the European Union. For this project, the software was extended by Erik Marchi in order to teach emotional expression to autistic children, based on automatic emotion recognition and visualization. In 2013, the company audEERING acquired the rights to the code-base from the Technical University of Munich and version 2.0 was published under a source-available research license. Until 2016, openSMILE was downloaded more than 50,000 times worldwide and has established itself as a standard toolkit for emotion recognition. == Awards == openSMILE was awarded in 2010 in the context of the ACM Multimedia Open Source Competition. The software tool is applied in numerous scientific publications on automatic emotion recognition. openSMILE and its extension openEAR have been cited in more than 1000 scientific publications until today.
Read more →
Block swap algorithms

In computer algorithms, block swap algorithms swap two regions of elements of an array. It is simple to swap two non-overlapping regions of an array of equal size. However, it is not as simple to swap two contiguous regions of an array of unequal sizes (algorithms that perform such swapping are called rotation algorithms). A few well-known algorithms can accomplish this: Bentley's juggling (also known as the dolphin algorithm), Gries-Mills rotation, triple reversal algorithm, conjoined triple reversal algorithm (also known as the trinity rotation) and Successive rotation. == Triple reversal algorithm == The triple reversal algorithm is the simplest to explain, using rotations. A rotation is an in-place reversal of array elements. This method swaps two elements of an array from outside in within a range. The rotation works for an even or odd number of array elements. The reversal algorithm uses three in-place rotations to accomplish an in-place block swap: Rotate region A Rotate region B Rotate region AB Where A and B are adjacent regions of an array that together form the region AB. Gries-Mills and reversal algorithms perform better than Bentley's juggling, because of their cache-friendly memory access pattern behavior. The triple reversal algorithm parallelizes well, because rotations can be split into sub-regions, which can be rotated independently of others.
Read more →
Absher (application)

Absher (Arabic: أبشر ‘Absher, roughly meaning "good tidings" or "yes, done") is a smartphone application and web portal which allows citizens and residents of Saudi Arabia to use a variety of governmental services. Amongst several other services with the Absher app, it can be used to apply for jobs and Hajj permits, passport info can be updated, and electronic crimes can be reported. The application provides around 280 services for residents of Saudi Arabia including but not limited to making appointments, renewing passports, residents' cards, IDs, driver's licenses and others, and, controversially, enables Saudi men to track the whereabouts of women they control as part of the country's male guardianship system. The app can be downloaded from the Google Play Store and Apple App Store and is supplied by the Saudi Interior Ministry. According to the Ministry of the Interior, Absher has more than 20 million users. As of February 2019, Absher has been downloaded 4.2 million times from the App Store. Some services provided through Absher can also be accessed through the website absher.sa. In March 2021, Saudi Arabia launched the digital version of the Absher for individuals app through which the users can download a copy of their digital ID. Then, new services were added to the platform such as online birth and death registration services, requesting amendments to academic credentials, correcting names in English and marital status and requesting civil records of children. == Impact on women's rights == The app has been criticized by various human rights activists, human rights organisations and international communities. The US and European countries have also condemned the app and urged the kingdom to end its male guardianship system. Absher gained media attention in 2019 for its functions supporting the Saudi policy of male guardianship following an investigation by Business Insider. The app allows for designated guardians to receive notifications if a woman under their guardianship passes through an airport and subsequently gives them the option to withdraw her right to travel. In a few cases, this system has been circumvented by women who have been able to gain control over its settings and use it to allow themselves to travel. US Senator Ron Wyden of Oregon wrote a letter to the CEO's of Apple and Google, criticizing the app and demanding for its removal immediately. Wyden said "American companies should not enable or facilitate the Saudi government's patriarchy," and called the Saudi system of control over women "abhorrent". According to the EU lawmakers, current rules imposed over the women by the Saudi government make women “second-class citizens”. The lawmakers also asked the EU states to continue to build pressure on Riyadh so as to improve the conditions of women and human rights. Amnesty International and Human Rights Watch accused Apple and Google of helping "enforce gender apartheid" by hosting the app. US congresswomen Rep. Katherine Clark and Rep. Carolyn B. Maloney condemned the kingdom's male guardianship system that reflected from the app, calling Absher a "patriarchal weapon" and asking for its removal. In response to the criticism received by Absher, Apple chief executive officer Tim Cook stated in February 2019 that he intended to investigate the situation. Similarly, Google announced that it would also review the application. After a prompt review, Google declined to remove the app from Google Play, citing that it did not violate the agreed upon terms and conditions of the store. Saudi doctor Khawla Al-Kuraya supported this app an editorial in Bloomberg News. Kuraya wrote that Absher helped Saudi women avoid governmental bureaucracy as it allows their male guardians to process their travel permits anywhere and anytime through Absher. Although she believes that the guardianship system needs to be reconsidered, she thinks that Absher is an important step towards facilitating women-guardians related issues in Saudi Arabia. Absher manager Atiyah Al-Anazy announced in 2019 that two million women were using the application in Saudi Arabia to facilitate their transactions. Some female users stated that the application has made their movement and travel-related issues easier. New measures were introduced that year to allow Saudi women above the age of 18 to travel without their male guardians, which ultimately released male authoritative rights on women. A law was subsequently passed allowing women over the age of 21 to receive a passport and travel without prior male permission.
Read more →
Collective operation

Collective operations are building blocks for interaction patterns, that are often used in SPMD algorithms in the parallel programming context. Hence, there is an interest in efficient realizations of these operations. A realization of the collective operations is provided by the Message Passing Interface (MPI). == Definitions == In all asymptotic runtime functions, we denote the latency α {\displaystyle \alpha } (or startup time per message, independent of message size), the communication cost per word β {\displaystyle \beta } , the number of processing units p {\displaystyle p} and the input size per node n {\displaystyle n} . In cases where we have initial messages on more than one node we assume that all local messages are of the same size. To address individual processing units we use p i ∈ { p 0 , p 1 , … , p p − 1 } {\displaystyle p_{i}\in \{p_{0},p_{1},\dots ,p_{p-1}\}} . If we do not have an equal distribution, i.e. node p i {\displaystyle p_{i}} has a message of size n i {\displaystyle n_{i}} , we get an upper bound for the runtime by setting n = max ( n 0 , n 1 , … , n p − 1 ) {\displaystyle n=\max(n_{0},n_{1},\dots ,n_{p-1})} . A distributed memory model is assumed. The concepts are similar for the shared memory model. However, shared memory systems can provide hardware support for some operations like broadcast (§ Broadcast) for example, which allows convenient concurrent read. Thus, new algorithmic possibilities can become available. == Broadcast == The broadcast pattern is used to distribute data from one processing unit to all processing units, which is often needed in SPMD parallel programs to dispense input or global values. Broadcast can be interpreted as an inverse version of the reduce pattern (§ Reduce). Initially only root r {\displaystyle r} with i d {\displaystyle id} 0 {\displaystyle 0} stores message m {\displaystyle m} . During broadcast m {\displaystyle m} is sent to the remaining processing units, so that eventually m {\displaystyle m} is available to all processing units. Since an implementation by means of a sequential for-loop with p − 1 {\displaystyle p-1} iterations becomes a bottleneck, divide-and-conquer approaches are common. One possibility is to utilize a binomial tree structure with the requirement that p {\displaystyle p} has to be a power of two. When a processing unit is responsible for sending m {\displaystyle m} to processing units i . . j {\displaystyle i..j} , it sends m {\displaystyle m} to processing unit ⌈ ( i + j ) / 2 ⌉ {\displaystyle \left\lceil (i+j)/2\right\rceil } and delegates responsibility for the processing units ⌈ ( i + j ) / 2 ⌉ . . j {\displaystyle \left\lceil (i+j)/2\right\rceil ..j} to it, while its own responsibility is cut down to i . . ⌈ ( i + j ) / 2 ⌉ − 1 {\displaystyle i..\left\lceil (i+j)/2\right\rceil -1} . Binomial trees have a problem with long messages m {\displaystyle m} . The receiving unit of m {\displaystyle m} can only propagate the message to other units, after it received the whole message. In the meantime, the communication network is not utilized. Therefore pipelining on binary trees is used, where m {\displaystyle m} is split into an array of k {\displaystyle k} packets of size ⌈ n / k ⌉ {\displaystyle \left\lceil n/k\right\rceil } . The packets are then broadcast one after another, so that data is distributed fast in the communication network. Pipelined broadcast on balanced binary tree is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , whereas for the non-pipelined case it takes O ( ( α + β n ) log ⁡ p ) {\displaystyle {\mathcal {O}}((\alpha +\beta n)\log p)} cost. == Reduce == The reduce pattern is used to collect data or partial results from different processing units and to combine them into a global result by a chosen operator. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by ⊗ {\displaystyle \otimes } and the result is eventually stored on p 0 {\displaystyle p_{0}} . The reduction operator ⊗ {\displaystyle \otimes } must be associative at least. Some algorithms require a commutative operator with a neutral element. Operators like s u m {\displaystyle sum} , m i n {\displaystyle min} , m a x {\displaystyle max} are common. Implementation considerations are similar to broadcast (§ Broadcast). For pipelining on binary trees the message must be representable as a vector of smaller object for component-wise reduction. Pipelined reduce on a balanced binary tree is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} . == All-Reduce == The all-reduce pattern (also called allreduce) is used if the result of a reduce operation (§ Reduce) must be distributed to all processing units. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by an operator ⊗ {\displaystyle \otimes } and the result is eventually stored on all p i {\displaystyle p_{i}} . Analog to the reduce operation, the operator ⊗ {\displaystyle \otimes } must be at least associative. All-reduce can be interpreted as a reduce operation with a subsequent broadcast (§ Broadcast). For long messages a corresponding implementation is suitable, whereas for short messages, the latency can be reduced by using a hypercube (Hypercube (communication pattern) § All-Gather/ All-Reduce) topology, if p {\displaystyle p} is a power of two. All-reduce can also be implemented with a butterfly algorithm and achieve optimal latency and bandwidth. All-reduce is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , since reduce and broadcast are possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} with pipelining on balanced binary trees. All-reduce implemented with a butterfly algorithm achieves the same asymptotic runtime. == Prefix-Sum/Scan == The prefix-sum or scan operation is used to collect data or partial results from different processing units and to compute intermediate results by an operator, which are stored on those processing units. It can be seen as a generalization of the reduce operation (§ Reduce). Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} . The operator ⊗ {\displaystyle \otimes } must be at least associative, whereas some algorithms require also a commutative operator and a neutral element. Common operators are s u m {\displaystyle sum} , m i n {\displaystyle min} and m a x {\displaystyle max} . Eventually processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ <= i {\displaystyle \otimes _{i'<=i}} m i ′ {\displaystyle m_{i'}} . In the case of the so-called exclusive prefix sum, processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ < i {\displaystyle \otimes _{i' Read more →
AIVA

AIVA (Artificial Intelligence Virtual Artist) is an electronic composer recognized by the SACEM. == Description == Created in February 2016, AIVA specializes in classical and symphonic music composition. It became the world's first virtual composer to be recognized by a music society (SACEM). By reading a large collection of existing works of classical music (written by human composers such as Bach, Beethoven, Mozart) AIVA is capable of detecting regularities in music and on this base composing on its own. The algorithm AIVA is based on deep learning and reinforcement learning architectures. Since January 2019, the company offers a commercial product, Music Engine, capable of generating short (up to 3 minutes) compositions in various styles (rock, pop, jazz, fantasy, shanty, tango, 20th century cinematic, modern cinematic, and Chinese). AIVA was presented at TED by Pierre Barreau. == Discography == AIVA is a published composer; its first studio album "Genesis" was released in November 2016. Second album "Among the Stars" in 2018. 2016 CD album « Genesis » Hv-Com – LEPM 048427. Track listing "Genesis": 2018 CD album « Among the Stars » Hv-Com – LEPM 048708 Avignon Symphonic Orchestra [ORAP] also performed Aiva's compositions [2] in April 2017.
Read more →
Collective operation

Collective operations are building blocks for interaction patterns, that are often used in SPMD algorithms in the parallel programming context. Hence, there is an interest in efficient realizations of these operations. A realization of the collective operations is provided by the Message Passing Interface (MPI). == Definitions == In all asymptotic runtime functions, we denote the latency α {\displaystyle \alpha } (or startup time per message, independent of message size), the communication cost per word β {\displaystyle \beta } , the number of processing units p {\displaystyle p} and the input size per node n {\displaystyle n} . In cases where we have initial messages on more than one node we assume that all local messages are of the same size. To address individual processing units we use p i ∈ { p 0 , p 1 , … , p p − 1 } {\displaystyle p_{i}\in \{p_{0},p_{1},\dots ,p_{p-1}\}} . If we do not have an equal distribution, i.e. node p i {\displaystyle p_{i}} has a message of size n i {\displaystyle n_{i}} , we get an upper bound for the runtime by setting n = max ( n 0 , n 1 , … , n p − 1 ) {\displaystyle n=\max(n_{0},n_{1},\dots ,n_{p-1})} . A distributed memory model is assumed. The concepts are similar for the shared memory model. However, shared memory systems can provide hardware support for some operations like broadcast (§ Broadcast) for example, which allows convenient concurrent read. Thus, new algorithmic possibilities can become available. == Broadcast == The broadcast pattern is used to distribute data from one processing unit to all processing units, which is often needed in SPMD parallel programs to dispense input or global values. Broadcast can be interpreted as an inverse version of the reduce pattern (§ Reduce). Initially only root r {\displaystyle r} with i d {\displaystyle id} 0 {\displaystyle 0} stores message m {\displaystyle m} . During broadcast m {\displaystyle m} is sent to the remaining processing units, so that eventually m {\displaystyle m} is available to all processing units. Since an implementation by means of a sequential for-loop with p − 1 {\displaystyle p-1} iterations becomes a bottleneck, divide-and-conquer approaches are common. One possibility is to utilize a binomial tree structure with the requirement that p {\displaystyle p} has to be a power of two. When a processing unit is responsible for sending m {\displaystyle m} to processing units i . . j {\displaystyle i..j} , it sends m {\displaystyle m} to processing unit ⌈ ( i + j ) / 2 ⌉ {\displaystyle \left\lceil (i+j)/2\right\rceil } and delegates responsibility for the processing units ⌈ ( i + j ) / 2 ⌉ . . j {\displaystyle \left\lceil (i+j)/2\right\rceil ..j} to it, while its own responsibility is cut down to i . . ⌈ ( i + j ) / 2 ⌉ − 1 {\displaystyle i..\left\lceil (i+j)/2\right\rceil -1} . Binomial trees have a problem with long messages m {\displaystyle m} . The receiving unit of m {\displaystyle m} can only propagate the message to other units, after it received the whole message. In the meantime, the communication network is not utilized. Therefore pipelining on binary trees is used, where m {\displaystyle m} is split into an array of k {\displaystyle k} packets of size ⌈ n / k ⌉ {\displaystyle \left\lceil n/k\right\rceil } . The packets are then broadcast one after another, so that data is distributed fast in the communication network. Pipelined broadcast on balanced binary tree is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , whereas for the non-pipelined case it takes O ( ( α + β n ) log ⁡ p ) {\displaystyle {\mathcal {O}}((\alpha +\beta n)\log p)} cost. == Reduce == The reduce pattern is used to collect data or partial results from different processing units and to combine them into a global result by a chosen operator. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by ⊗ {\displaystyle \otimes } and the result is eventually stored on p 0 {\displaystyle p_{0}} . The reduction operator ⊗ {\displaystyle \otimes } must be associative at least. Some algorithms require a commutative operator with a neutral element. Operators like s u m {\displaystyle sum} , m i n {\displaystyle min} , m a x {\displaystyle max} are common. Implementation considerations are similar to broadcast (§ Broadcast). For pipelining on binary trees the message must be representable as a vector of smaller object for component-wise reduction. Pipelined reduce on a balanced binary tree is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} . == All-Reduce == The all-reduce pattern (also called allreduce) is used if the result of a reduce operation (§ Reduce) must be distributed to all processing units. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by an operator ⊗ {\displaystyle \otimes } and the result is eventually stored on all p i {\displaystyle p_{i}} . Analog to the reduce operation, the operator ⊗ {\displaystyle \otimes } must be at least associative. All-reduce can be interpreted as a reduce operation with a subsequent broadcast (§ Broadcast). For long messages a corresponding implementation is suitable, whereas for short messages, the latency can be reduced by using a hypercube (Hypercube (communication pattern) § All-Gather/ All-Reduce) topology, if p {\displaystyle p} is a power of two. All-reduce can also be implemented with a butterfly algorithm and achieve optimal latency and bandwidth. All-reduce is possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , since reduce and broadcast are possible in O ( α log ⁡ p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} with pipelining on balanced binary trees. All-reduce implemented with a butterfly algorithm achieves the same asymptotic runtime. == Prefix-Sum/Scan == The prefix-sum or scan operation is used to collect data or partial results from different processing units and to compute intermediate results by an operator, which are stored on those processing units. It can be seen as a generalization of the reduce operation (§ Reduce). Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} . The operator ⊗ {\displaystyle \otimes } must be at least associative, whereas some algorithms require also a commutative operator and a neutral element. Common operators are s u m {\displaystyle sum} , m i n {\displaystyle min} and m a x {\displaystyle max} . Eventually processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ <= i {\displaystyle \otimes _{i'<=i}} m i ′ {\displaystyle m_{i'}} . In the case of the so-called exclusive prefix sum, processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ < i {\displaystyle \otimes _{i' Read more →
LMArena

Arena (formerly LMArena and Chatbot Arena) is a public, web-based platform that evaluates large language models (LLMs). Users enter prompts for two anonymous models to respond to and vote on the model that gave the better response, after which the models' identities are revealed. Users can also choose models to test themselves via the "Direct" selection. Companies which have supplied the company with their large language models include OpenAI, Google DeepMind, and Anthropic. The website has been used for preview releases of upcoming models. Chinese company DeepSeek tested its prototype models in the Arena months before its R1 model gained attention in Western media. Other notable pre-release models include OpenAI's GPT-5 under the codename "summit" and Google DeepMind's Gemini 2.5 Flash Image (an image-generation and editing model) under the codename "Nano Banana". Research has identified specific limitations in Arena's methodology. == History == Chatbot Arena was released on April 24, 2023. In June 2024, Chatbot Arena added image support. In September 2024, Chatbot Arena moved to its own dedicated domain name, lmarena.ai (or LMArena). In April 2025, Meta released Llama 4. Llama 4 Maverick beat GPT-4o and Gemini 2.0 Flash on LMArena, but the version of Maverick on LMArena unfairly differed from the publicly available version. LMArena updated their policies in response. In April 2025, LMArena incorporated as an independent company. That May, LMArena raised $100 million in a seed funding round, valuing the company at $600 million. Participants in the seed funding round included Andreessen Horowitz, UC Investments, Lightspeed Venture Partners, Felicis Ventures, and Kleiner Perkins. On January 6, 2026, LMArena announced the closing of a $150 million Series A funding round, bringing the company’s post-money valuation to approximately $1.7 billion. The round was led by Felicis and UC Investments (University of California), with participation from Andreessen Horowitz, The House Fund, LDVP, Kleiner Perkins, Lightspeed Venture Partners, and Laude Ventures. In January 2026, LMArena added video support. On January 28, 2026, LMArena rebranded to "Arena".
Read more →
Manhattan address algorithm

The Manhattan address algorithm is a series of formulas used to estimate the closest east–west cross street for building numbers on north–south avenues in the New York City borough of Manhattan. == Algorithm == To find the approximate number of the closest cross street, divide the building number by a divisor (generally 20) and add (or subtract) the "tricky number" from the table below: For the north–south avenues, there are typically 20 address numbers between consecutive east–west streets (10 on either side of the avenue). A standard land lot on each avenue was originally 20 feet (6.1 m) wide, and there is about 200 feet (61 m) between each pair of east–west streets, for ten land lots between each pair of streets. The exceptions are Riverside Drive, as well as Fifth Avenue and Central Park West between 59th and 110th streets, which use a divisor of 10. These avenues all have buildings only on one side of the street, with a park on the other side. The "tricky number" often corresponds to a street near the southern end of the avenue. There are some notable exceptions: York Avenue address numbers are continuations of Avenue A address numbers, since the avenue was originally called Avenue A. East End Avenue address numbers are continuations of Avenue B address numbers, since the avenue was originally called Avenue B. Sixth Avenue and Broadway start south of Houston Street, the southern boundary of the Manhattan street numbering system. Although Park Avenue's southern terminus is at 32nd Street, a homeowner at 34th Street wanted the address "1 Park Avenue" (this was later changed). === Examples === For example, if you are at 62 Avenue B, 62 ÷ 20 ≈ 3 {\displaystyle 62\div 20\approx 3} , then add the "tricky number" 3 {\displaystyle 3} to give 6 {\displaystyle 6} . The nearest cross street to 62 Avenue B is East 6th Street. If you are at 78 Riverside Drive, 78 ÷ 10 ≈ 8 {\displaystyle 78\div 10\approx 8} , then add the "tricky number" 72 {\displaystyle 72} to give 80 {\displaystyle 80} . The nearest cross street to 78 Riverside Drive is West 80th Street. If you are at 501 5th Avenue, 501 ÷ 20 ≈ 25 {\displaystyle 501\div 20\approx 25} , then add the "tricky number" 18 {\displaystyle 18} to give 43 {\displaystyle 43} . The nearest cross street to 501 5th Avenue is actually 42nd Street, not 43rd Street, as the Manhattan address algorithm only gives approximate answers.
Read more →
Ordered key–value store

An ordered key–value store (OKVS) is a type of data storage paradigm that can support multi-model databases. An OKVS is an ordered mapping of bytes to bytes. An OKVS will keep the key–value pairs sorted by the key lexicographic order. OKVS systems provides different set of features and performance trade-offs. Most of them are shipped as a library without network interfaces, in order to be embedded in another process. Most OKVS support ACID guarantees. Some OKVS are distributed databases. Ordered key–value stores found their way into many modern database systems including NewSQL database systems. == History == The origin of ordered key–value store stems from the work of Ken Thompson on dbm in 1979. Later in 1991, Berkeley DB was released that featured a B-Tree backend that allowed the keys to stay sorted. Berkeley DB was said to be very fast and made its way into various commercial product. It was included in Python standard library until 2.7. In 2009, Tokyo Cabinet was released that was superseded by Kyoto Cabinet that support both transaction and ordered keys. In 2011, LMDB was created to replace Berkeley DB in OpenLDAP. There is also Google's LevelDB that was forked by Facebook in 2012 as RocksDB. In 2014, WiredTiger, successor of Berkeley DB was acquired by MongoDB and is since 2019 the primary backend of MongoDB database. Other notable implementation of the OKVS paradigm are Sophia and SQLite3 LSM extension. Another notable use of OKVS paradigm is the multi-model database system called ArangoDB based on RocksDB. Some NewSQL databases are supported by ordered key–value stores. JanusGraph, a property graph database, has both a Berkeley DB backend and FoundationDB backend. == Key concepts == === Lexicographic encoding === There are algorithms that encode basic data types (boolean, string, number) and composition of those data types inside sorted containers (tuple, list, vector) that preserve their natural ordering. It is possible to work with an ordered key–value store without having to work directly with bytes. In FoundationDB, it is called the tuple layer. === Range query === Inside an OKVS, keys are ordered, and because of that it is possible to do range queries. A range query retrieves all keys between two specified keys, ensuring that the fetched keys are returned in a sorted order. === Subspaces === === Key composition === One can construct key spaces to build higher level abstractions. The idea is to construct keys, that takes advantage of the ordered nature of the top level key space. When taking advantage of the ordered nature of the key space, one can query ranges of keys that have particular pattern. === Denormalization === Denormalization, as in, repeating the same piece of data in multiple subspace is common practice. It allows to create secondary representation, also called indices, that will allow to speed up queries. == Higher level abstractions == The following abstraction or databases were built on top ordered key–value stores: Timeseries database, Record Database, also known as Row store databases, they behave similarly to what is dubbed RDBMS, Tuple Stores, also known as Triple Store or Quad Store but also Generic Tuple Store, Document database, that mimics MongoDB API, Full-text search Geographic Information Systems Property Graph Versioned Data Vector space database for Approximate Nearest Neighbor All those abstraction can co-exist with the same OKVS database and when ACID is supported, the operations happens with the guarantees offered by the transaction system. == Feature matrix == == Use-cases == OKVS are useful to implement two strategies: optimize a small feature e.g. to make a 10% improvement in read or write latency; the second strategy is to take advantage of the distributed nature of FoundationDB, and TiKV, for which there is no equivalent at very large scale in resilience. Both users need to re-implement the needed high level abstractions, because there are no portable ready-to-use libraries of high-level abstraction. There is still a complex balance, of complexity, maintainability, fine-tuning, and readily available features that makes it still a choice of experts. Sometime more specialized data-structures can be faster than a high-level abstraction on top of an OKVS. Another interest of OKVS paradigm stems from it simple, and versatile interface, that makes it an interesting target for experimental storage algorithms, and data structures.
Read more →