Lorraine Susan (Lori) Levin is an American computer scientist and computational linguist specializing in natural language processing, particularly involving syntax, morphosyntax, and languages with small corpora. She is a research professor in the Language Technologies Institute of the Carnegie Mellon University School of Computer Science, and one of the founders of the North American Computational Linguistics Open Competition. == Education and career == Levin has a 1979 bachelor's degree in linguistics (summa cum laude) from the University of Pennsylvania, and a 1986 Ph.D. in linguistics from the Massachusetts Institute of Technology. Her dissertation, Operations on Lexical Forms: Unaccusative Rules in Germanic Languages, was jointly supervised by Joan Bresnan and Kenneth L. Hale. She worked as an assistant professor of linguistics at the University of Pittsburgh from 1983 until 1988, when she joined the Carnegie Mellon University Language Technologies Institute. == Recognition == Levin was named as a Fellow of the Association for Computational Linguistics in 2025, "for pioneering work on the use of phonetics, syntax, lexical semantics and dialogue modeling in machine translation and in the transfer of NLP technologies to low resource languages, as well as an enduring contribution to the North American Computational Linguistics Olympiad". Levin was awarded the Antonio Zampolli prize of the ELRA Language Resources Association at the LREC 2026 conference.
Deep Learning Anti-Aliasing
Deep Learning Anti-Aliasing (DLAA) is a form of spatial anti-aliasing developed by Nvidia. DLAA depends on and requires Tensor Cores available in Nvidia RTX cards. DLAA is similar to Deep Learning Super Sampling (DLSS) in its anti-aliasing method, with one important differentiation being that the goal of DLSS is to increase performance at the cost of image quality, whereas the main priority of DLAA is improving image quality at the cost of performance (irrelevant of resolution upscaling or downscaling). DLAA is similar to temporal anti-aliasing (TAA) in that they are both spatial anti-aliasing solutions relying on past frame data. Compared to TAA, DLAA is substantially better when it comes to shimmering, flickering, and handling small meshes like wires. == Technical overview == DLAA collects game rendering data including raw low-resolution input, motion vectors, depth buffers, and exposure information. This information feeds into a convolutional neural network that processes the image to reduce aliasing while preserving fine detail. The neural network architecture employs an auto-encoder design trained on high-quality reference images. The training dataset includes diverse scenarios focusing on challenging cases like sub-pixel details, high-contrast edges, and transparent surfaces. The network then processes frames in real-time. Unlike traditional anti-aliasing solutions that rely on manually written heuristics, such as TAA, DLAA uses its neural network to preserve fine details while eliminating unwanted visual artifacts. == History == DLAA was initially called and marketed by Nvidia as DLSS 2x. The first game that added support for DLAA was The Elder Scrolls Online, which implemented the feature in 2021. By June 2022, DLAA was only available in six games. This number rose to 17 by February 2023. In June 2023, TechPowerUp reported that "DLAA is seeing sluggish adoption among game developers", and that Nvidia was working on adding DLAA to the quality presets of DLSS to boost adoption. By December 2023, DLAA was supported in 41 games. In early 2025, an update for the Nvidia App added a driver-based DLSS override feature that enables users to activate DLAA even in games that do not support it natively. == Differences between TAA and DLAA == TAA is used in many modern video games and game engines; however, all previous implementations have used some form of manually written heuristics to prevent temporal artifacts such as ghosting and flickering. One example of this is neighborhood clamping which forcefully prevents samples collected in previous frames from deviating too much compared to nearby pixels in newer frames. This helps to identify and fix many temporal artifacts, but deliberately removing fine details in this way is analogous to applying a blur filter, and thus the final image can appear blurry when using this method. DLAA uses an auto-encoder convolutional neural network trained to identify and fix temporal artifacts, instead of manually programmed heuristics as mentioned above. Because of this, DLAA can generally resolve detail better than other TAA and TAAU implementations, while also removing most temporal artifacts. == Differences between DLSS and DLAA == While DLSS handles upscaling with a focus on performance, DLAA handles anti-aliasing with a focus on visual quality. DLAA runs at the given screen resolution with no upscaling or downscaling functionality provided by DLAA. DLSS and DLAA share the same AI-driven anti-aliasing method. As such, DLAA functions like DLSS without the upscaling part. Both are made by Nvidia and require Tensor Cores. However, DLSS and DLAA cannot be enabled at the same time, only one can be selected depending on whether performance or image quality is prioritized. == Reception == TechPowerUp found that "[c]ompared to TAA and DLSS, DLAA is clearly producing the best image quality, especially at lower resolutions", arguing that, while "DLSS was already doing a better job than TAA at reconstructing small objects", "DLAA does an even better job". In a Cyberpunk 2077 performance test, IGN stated that "DLAA provided somewhat similar results [FPS wise] to the normal raster mode in most cases but got significant performance boost with the help of frame generation", a feature not available when using native resolution. Rock Paper Shotgun noted that, while DLAA is "not a completely perfect form of anti-aliasing, as the occasional jaggies are present", it "looks a lot sharper overall [than TAA], and especially in motion." According to PC World, "DLAA offers very good anti-aliasing without losing visual information — alternatives like TAA tend to struggle during motion-filled scenes, where DLAA doesn’t. Furthermore, DLAA’s loss of performance is lower than with conventional anti-aliasing methods."
Feistel cipher
In cryptography, a Feistel cipher (also known as Luby–Rackoff block cipher) is a symmetric structure used in the construction of block ciphers, named after the German-born physicist and cryptographer Horst Feistel, who did pioneering research while working for IBM; it is also commonly known as a Feistel network. A large number of block ciphers use the scheme, including the US Data Encryption Standard, the Soviet/Russian GOST (aka Magma) and the more recent Blowfish and Twofish ciphers. In a Feistel cipher, encryption and decryption are very similar operations, and both consist of iteratively running a function called a "round function" a fixed number of times. == History == Many modern symmetric block ciphers are based on Feistel networks. Feistel networks were first seen commercially in IBM's Lucifer cipher, designed by Horst Feistel and Don Coppersmith in 1973. Feistel networks gained respectability when the U.S. Federal Government adopted the DES (a cipher based on Lucifer, with changes made by the NSA) in 1976. Like other components of the DES, the iterative nature of the Feistel construction makes implementing the cryptosystem in hardware easier (particularly on the hardware available at the time of DES's design). == Design == A Feistel network uses a round function, a function which takes two inputs – a data block and a subkey – and returns one output of the same size as the data block. In each round, the round function is run on half of the data to be encrypted, and its output is XORed with the other half of the data. This is repeated a fixed number of times, and the final output is the encrypted data. An important advantage of Feistel networks compared to other cipher designs such as substitution–permutation networks (SP-networks) is that the entire operation is guaranteed to be invertible (that is, encrypted data can be decrypted), even if the round function is not itself invertible. The round function can be made arbitrarily complicated, since it does not need to be designed to be invertible. Furthermore, the encryption and decryption operations are very similar, even identical in some cases, requiring only a reversal of the key schedule. Therefore, the size of the code or circuitry required to implement such a cipher is nearly halved. Unlike SP-networks, Feistel networks also do not depend on a substitution box that could cause timing side-channels in software implementations. == Theoretical work == The structure and properties of Feistel ciphers have been extensively analyzed by cryptographers. Michael Luby and Charles Rackoff analyzed the Feistel cipher construction and proved that if the round function is a cryptographically secure pseudorandom function, with Ki used as the seed, then 3 rounds are sufficient to make the block cipher a pseudorandom permutation, while 4 rounds are sufficient to make it a "strong" pseudorandom permutation (which means that it remains pseudorandom even to an adversary who gets oracle access to its inverse permutation). Because of this very important result of Luby and Rackoff, Feistel ciphers are sometimes called Luby–Rackoff block ciphers. Further theoretical work has generalized the construction somewhat and given more precise bounds for security. == Construction details == Let F {\displaystyle \mathrm {F} } be the round function and let K 0 , K 1 , … , K n {\displaystyle K_{0},K_{1},\ldots ,K_{n}} be the sub-keys for the rounds 0 , 1 , … , n {\displaystyle 0,1,\ldots ,n} respectively. Then the basic operation is as follows: Split the plaintext block into two equal pieces: ( L 0 {\displaystyle L_{0}} , R 0 {\displaystyle R_{0}} ). For each round i = 0 , 1 , … , n {\displaystyle i=0,1,\dots ,n} , compute L i + 1 = R i , {\displaystyle L_{i+1}=R_{i},} R i + 1 = L i ⊕ F ( R i , K i ) , {\displaystyle R_{i+1}=L_{i}\oplus \mathrm {F} (R_{i},K_{i}),} where ⊕ {\displaystyle \oplus } means XOR. Then the ciphertext is ( R n + 1 , L n + 1 ) {\displaystyle (R_{n+1},L_{n+1})} . Decryption of a ciphertext ( R n + 1 , L n + 1 ) {\displaystyle (R_{n+1},L_{n+1})} is accomplished by computing for i = n , n − 1 , … , 0 {\displaystyle i=n,n-1,\ldots ,0} R i = L i + 1 , {\displaystyle R_{i}=L_{i+1},} L i = R i + 1 ⊕ F ( L i + 1 , K i ) . {\displaystyle L_{i}=R_{i+1}\oplus \operatorname {F} (L_{i+1},K_{i}).} Then ( L 0 , R 0 ) {\displaystyle (L_{0},R_{0})} is the plaintext again. The diagram illustrates both encryption and decryption. Note the reversal of the subkey order for decryption; this is the only difference between encryption and decryption. === Unbalanced Feistel cipher === Unbalanced Feistel ciphers use a modified structure where L 0 {\displaystyle L_{0}} and R 0 {\displaystyle R_{0}} are not of equal lengths. The Skipjack cipher is an example of such a cipher. The Texas Instruments digital signature transponder uses a proprietary unbalanced Feistel cipher to perform challenge–response authentication. The Thorp shuffle is an extreme case of an unbalanced Feistel cipher in which one side is a single bit. This has better provable security than a balanced Feistel cipher but requires more rounds. There exists Type-1, Type-2, and Type-3 Feistel networks, where the Feistel function is one fourth the size of the block but operates a varying number of times within one round. === Other uses === The Feistel construction is also used in cryptographic algorithms other than block ciphers. For example, the optimal asymmetric encryption padding (OAEP) scheme uses a simple Feistel network to randomize ciphertexts in certain asymmetric-key encryption schemes. A generalized Feistel algorithm can be used to create strong permutations on small domains of size not a power of two (see format-preserving encryption). === Feistel networks as a design component === Whether the entire cipher is a Feistel cipher or not, Feistel-like networks can be used as a component of a cipher's design. For example, MISTY1 is a Feistel cipher using a three-round Feistel network in its round function, Skipjack is a modified Feistel cipher using a Feistel network in its G permutation, and Threefish (part of Skein) is a non-Feistel block cipher that uses a Feistel-like MIX function. == List of Feistel ciphers == Feistel or modified Feistel: Generalised Feistel: CAST-256 CLEFIA MacGuffin RC2 RC6 Skipjack SMS4
Virtual influencer
A virtual influencer, sometimes described as a virtual persona or virtual model, is a computer-generated fictional character that can be used for a variety of marketing-related purposes, but most frequently for social media marketing, in lieu of online human "influencers". Most virtual influencers are designed using computer graphics and motion capture technology to resemble real people in realistic situations. Common derivatives of virtual influencers include VTubers, which broadly refer to online entertainers and YouTubers who represent themselves using virtual avatars instead of their physical selves. == History == Virtual influencers are fundamentally synonymous with virtual idols, which originate from Japan's anime and Japanese idol culture that dates back to the 1980s. The first virtual idol created was Lynn Minmay, a fictional singer and main character of the anime television series Super Dimension Fortress Macross (1982) and the animated film adaptation Macross: Do You Remember Love? (1984). Minmay's success led to the production of more Japanese virtual idols, such as EVE from the Japanese cyberpunk anime Megazone 23 (1985), and Sharon Apple in Macross Plus (1994). Virtual idols were not always well received – in 1995, Japanese talent agency Horipro created Kyoko Date, which was inspired by the Macross franchise and dating sim games such as Tokimeki Memorial (1994). Date failed to gain commercial success despite drawing headlines for her debut as a CGI idol, largely due to technical limitations leading to issues such as unnatural movements, an issue also known as the uncanny valley. Since their inception, many virtual idols created have achieved continual success, with notable names including the Vocaloid singer Hatsune Miku, and the VTuber Kizuna AI. Technological advancements have also enabled production teams to use artificial intelligence and advanced techniques to customize the personalities and behavior of virtual idols. Due to modern-day advancements in technology, many virtual idols have held real-life tours and events. Notable ones include Hatsune Miku's titular tour Miku Expo and Hololive's concerts with many of their idols from their English, Japanese and Indonesian branches. Some notable events including virtual singers and influencers have included: Hatsune Miku opening for Lady Gaga in 2014 and Hoshimachi Suisei's concerts at the famous Budokan venue in Japan and her addition to the Forbes Japan list of '30 Under 30' individuals who are changing the world in their respective fields. == Benefits and criticism == From a branding perspective, virtual influencers are perceived to be much less likely to be mired in scandals. In China, celebrities caught in bad publicity such as singer Wang Leehom and entertainer Kris Wu have heightened the appeal of virtual influencers, since their existence relies entirely on computer-generated imagery and they are therefore unlikely to cause any damage to a brand's image by association. Some studies have also suggested that Generation Z consumers have a unique appetite for virtual idols and influencers, since they grew up in the age of the internet. Studies also show that human-like appearance of virtual influencers show higher message credibility than anime-like virtual influencers. Scholars and commentators have also questioned the ethics and cultural impact of virtual influencers, arguing that computer-generated personas can entrench unrealistic beauty standards while diffusing accountability for labor, identity, and consent. Business and marketing analysts have also warned that disclosure and governance remain inconsistent, recommending clearer guardrails and transparency when brands deploy synthetic spokespeople. In 2025, reporting highlighted concerns that AI-driven "virtual humans" could displace human creators and sales workers, intensifying debates over the future of creative labor and authenticity online. == Notable examples == === Virtual bands === Eternity - A South Korean virtual idol group formed by Pulse9. Gorillaz - A virtual band formed in 1998. K/DA - A virtual K-pop girl group created as part of the League of Legends video game franchise. MAVE: - A South Korean virtual girl group formed in 2023 by Metaverse Entertainment. Pentakill - A virtual heavy metal band created as part of the League of Legends video game franchise. Plave (band) - A South Korean virtual boy band formed by VLast. Squid Sisters and Off the Hook - Two virtual pop idol duos as part of the Splatoon series. Studio Killers - A Finnish-Danish-British virtual band formed in 2011. === Vocaloids === Hatsune Miku (modeled after Saki Fujita) Kagamine Rin/Len (modeled after Asami Shimoda) Megurine Luka (modeled after Yū Asakawa) Meiko (modeled after Meiko Haigō) Kaito (modeled after Naoto Fūga) === VTubers === Kano Kizuna AI Neuro-sama VShojo Ironmouse Projekt Melody Nijisanji Hololive Akai Haato Gawr Gura Hoshimachi Suisei Natsuiro Matsuri === Other examples === Ami Yamato Crazy Frog FN Meka IA Kuki AI Kyoko Date Kyra Miquela Naevis Shudu Gram
Kruskal count
The Kruskal count (also known as Kruskal's principle, Dynkin–Kruskal count, Dynkin's counting trick, Dynkin's card trick, coupling card trick or shift coupling) is a probabilistic concept originally demonstrated by the Russian mathematician Evgenii Borisovich Dynkin in the 1950s or 1960s discussing coupling effects and rediscovered as a card trick by the American mathematician Martin David Kruskal in the early 1970s as a side-product while working on another problem. It was published by Kruskal's friend Martin Gardner and magician Karl Fulves in 1975. This is related to a similar trick published by magician Alexander F. Kraus in 1957 as Sum total and later called Kraus principle. Besides uses as a card trick, the underlying phenomenon has applications in cryptography, code breaking, software tamper protection, code self-synchronization, control-flow resynchronization, design of variable-length codes and variable-length instruction sets, web navigation, object alignment, and others. == Card trick == The trick is performed with cards, but is more a magical-looking effect than a conventional magic trick. The magician has no access to the cards, which are manipulated by members of the audience. Thus sleight of hand is not possible. Rather the effect is based on the mathematical fact that the output of a Markov chain, under certain conditions, is typically independent of the input. A simplified version using the hands of a clock performed by David Copperfield is as follows. A volunteer picks a number from one to twelve and does not reveal it to the magician. The volunteer is instructed to start from 12 on the clock and move clockwise by a number of spaces equal to the number of letters that the chosen number has when spelled out. This is then repeated, moving by the number of letters in the new number. The output after three or more moves does not depend on the initially chosen number and therefore the magician can predict it.
Web development
Web development is the process of designing, developing and maintaining websites and web apps. Web development encompasses several different fields, most commonly referring to the programming of websites. Front-end development is the act of developing the user interface and client-side code, while back-end development focuses on the infrastructure behind a website, mainly server-side code. Since the World Wide Web was released publicly in 1993, web development has evolved greatly, with websites changing from a collection of static HTML pages to complex projects using frameworks, servers, and databases. == Overview == Web development includes many individual tasks, including web design, web content development, networking, and coding. Among web professionals, "web development" usually refers to the main non-design aspects of building websites: writing markup and coding. Web development is generally split into two fields: front-end development and back-end development. Front-end developers create the user interface of websites, turning web designs into HTML, CSS, and JavaScript code. Front-end developers must also make sure that websites work consistently across different browsers and devices. Back-end development, also known as server-side development, focuses on the infrastructure behind a website, including APIs, database management, and security. Some choose to be full-stack developers, meaning they work on both the front-end and back-end. == History == The World Wide Web is often categorised into three generations: Web 1.0, Web 2.0, and Web 3.0 (or Web3). It was invented in 1989, and released to the public in 1993. In the early years of the web, restrospecitvely referred to as Web 1.0, websites were simply a collection of static HTML files, and had limited interactivity. After the introduction of JavaScript in 1995, websites could contain logic, allowing for interactivity. The following year CSS was released, allowing greater control over the styling of web pages. In 1999, the term Web 2.0 was coined by Darcy DiNucci. The term later resurfaced in the early 2000s, as websites started to increase in complexity, requiring server-side services in addition to JavaScript. This led to the emergence of various new programming languages and frameworks designed for backend services, such as PHP, Active Server Pages, and Jakarta Server Pages. This enabled websites to do additional server-side processing, such as accessing databases. Another shift in web development was the release of the iPhone in 2007. This created a new medium for accessing the web, requiring a new approach to web development, and resulting in responsive web design, which allows a single website to appear different depending on the device running it. Later, progressive web apps were introduced, allowing websites to be installed on a device as an independent application. In the 2010s, JavaScript frameworks began to emerge, creating new ways to manipulate web pages, and increasing compatibility between web browsers. JQuery was popular in the early 2010s, but was later surpassed by other frameworks such as React and Vue.js. In the mid 2020s, use of AI became prevalent among web developers, with the 2025 Stack Overflow survey showing over 80% of developers saying the use AI at least monthly in their development process.
Data lake
A data lake is a system or repository of data stored in its natural/raw format, usually object blobs or files. A data lake is usually a single store of data including raw copies of source system data, sensor data, social data etc., and transformed data used for tasks such as reporting, visualization, advanced analytics, and machine learning. A data lake can include structured data from relational databases (rows and columns), semi-structured data (CSV, logs, XML, JSON), unstructured data (emails, documents, PDFs), and binary data (images, audio, video). A data lake can be established on premises (within an organization's data centers) or in the cloud (using cloud services). == Background == James Dixon, then chief technology officer at Pentaho, coined the term by 2011 to contrast it with data mart, which is a smaller repository of interesting attributes derived from raw data. In promoting data lakes, he argued that data marts have several inherent problems, such as information siloing. PricewaterhouseCoopers (PwC) said that data lakes could "put an end to data silos". In their study on data lakes, they noted that enterprises were "starting to extract and place data for analytics into a single, Hadoop-based repository." == Examples == Many companies use cloud storage services such as Google Cloud Storage and Amazon S3 or a distributed file system such as Apache Hadoop distributed file system (HDFS). There is a gradual academic interest in the concept of data lakes. For example, Personal DataLake at Cardiff University is a new type of data lake which aims at managing big data of individual users by providing a single point of collecting, organizing, and sharing personal data. Early data lakes, such as Hadoop 1.0, had limited capabilities because it only supported batch-oriented processing (Map Reduce). Interacting with it required expertise in Java, map reduce and higher-level tools like Apache Pig, Apache Spark and Apache Hive (which were also originally batch-oriented). == Criticism == Poorly managed data lakes have been facetiously called data swamps. In June 2015, David Needle characterized "so-called data lakes" as "one of the more controversial ways to manage big data". PwC was also careful to note in their research that not all data lake initiatives are successful. They quote Sean Martin, CTO of Cambridge Semantics: We see customers creating big data graveyards, dumping everything into Hadoop distributed file system (HDFS) and hoping to do something with it down the road. But then they just lose track of what’s there. The main challenge is not creating a data lake, but taking advantage of the opportunities it presents. They describe companies that build successful data lakes as gradually maturing their lake as they figure out which data and metadata are important to the organization. Another criticism is that the term data lake is used with many different meanings. It may be used to refer to, for example: any tools or data management practices that are not data warehouses; a particular technology for implementation; a raw data reservoir; a hub for ETL offload; or a central hub for self-service analytics. While critiques of data lakes are warranted, in many cases they apply to other data projects as well. For example, the definition of data warehouse is also changeable, and not all data warehouse efforts have been successful. In response to various critiques, McKinsey noted that the data lake should be viewed as a service model for delivering business value within the enterprise, not a technology outcome. == Data lakehouses == Data lakehouses are a hybrid approach that can ingest a variety of raw data formats like a data lake, while also providing ACID transactions and enforced data quality like a data warehouse.