Boba liberal is a term mostly used within the Asian diaspora communities in the West, especially in the United States. It describes someone of East or Southeast Asian descent living in the West who has a shallow, surface-level liberal outlook. It is also occasionally used to describe conservatives who weaponize their East or Southeast Asian identity. The neologism emerged among the Asian American leftist community on Twitter who accused "boba liberals" of only holding their liberal beliefs to appear more white-adjacent by engaging in progressive social movements or viewpoints, while at the same time disregarding and trivializing issues concerning Asians. Mary Chao, writing for The North Jersey Record, said that "Asians call peers boba liberals when they aspire to liberal whiteness." An article in The Yale Herald described it as a term "used to describe the ethnocentric politics of Asian Americans, usually of East Asian descent, who exclusively advocate for issues that benefit themselves, without acknowledging problematic dimensions of their own history and working to support other people of color." The feminist magazine Fem said that "the faces of boba liberalism are Asian Americans that are part of the middle and upper economic class. As a result, boba liberals disregard the negative effects of capitalism because they profit from it. For instance, boba liberals tend to focus on advocating for Asian representation in white spaces, or discussing whether or not wearing chopsticks in one's hair is culture appropriation. These topics are popular within boba liberal circles, all while dialogue regarding inequality, globalization, and racial injustice are purposely neglected." UnHerd notes that conservative Asian Americans have used the term not to critique capitalism, but to "aim at a small but influential group of progressive Asian-American activists who are supposedly selling out other Asians, especially working-class Asians, in order to win brownie points from elite, generally white liberals." MRAsians have similarly used the term to attack Asian American feminists who supported the Black Lives Matter movement. The Asian identity of boba liberals has often been accused of being shallow and superficial. Boba liberals are accused of using surface-level stereotypical Asian traits such as liking boba tea to bolster their Asian credentials. Plan A Magazine, an Asian diaspora magazine, described the film Crazy Rich Asians and the sitcom Fresh Off the Boat as "boba liberal media", calling them the result of "a specific kind of atomized identity politics". Other media outlets have connected the Crazy Rich Asians film to boba liberalism. == Controversy == The term "boba liberal" was coined in 2019 by Vietnamese American Twitter user Redmond (@diaspora_is_red) to analyze a form of Asian American liberalism through a Marxist lens. Redmond has criticized the misappropriation of their neologism by stripping away the Marxist framework by failing to discuss "socialism, communism, the capitalist system, imperialism, and the diaspora bourgeoisie" and conflating "boba liberalism" with the flawed concept of "East Asian privilege". In 2024, Redmond criticized misuse of the term by conservatives and liberals, and said "The term boba liberalism can go away for all I care. It's corny and stale". === United States === One commentator described boba liberals as supporting policies that primarily benefit upper-income Asian-Americans, and not necessarily the Asian-American community as a whole. Therefore, while the word "liberal" is used in the term, it is not mutually exclusive to one specific ideology, as it may also extend to conservative-aligned Asians in some areas, as they would often take advantage of the "model minority" label by defending such measures.
Pixel aspect ratio
A pixel aspect ratio (PAR) is a mathematical ratio that describes how the width of a pixel in a digital image compares to the height of that pixel. Most digital imaging systems display an image as a grid of tiny, square pixels. However, some imaging systems, especially those that must be compatible with standard-definition television motion pictures, display an image as a grid of rectangular pixels, in which the pixel width and height are different. Pixel aspect ratio describes this difference. Use of pixel aspect ratio mostly involves pictures pertaining to standard-definition television and some other exceptional cases. Most other imaging systems, including those that comply with SMPTE standards and practices, use square pixels. PAR is also known as sample aspect ratio and abbreviated SAR, though it can be confused with storage aspect ratio. == Introduction == The ratio of the width to the height of an image is known as the aspect ratio, or more precisely the display aspect ratio (DAR) – the aspect ratio of the image as displayed; for TV, DAR was traditionally 4:3 (a.k.a. fullscreen), with 16:9 (a.k.a. widescreen) now the standard for HDTV. In digital images, there is a distinction with the storage aspect ratio (SAR), which is the ratio of pixel dimensions. If an image is displayed with square pixels, then these ratios agree; if not, then non-square, "rectangular" pixels are used, and these ratios disagree. The aspect ratio of the pixels themselves is known as the pixel aspect ratio (PAR) – for square pixels this is 1:1 – and these are related by the identity: Rearranging (solving for PAR) yields: For example: A 640 × 480 VGA image has a SAR of 640/480 = 4:3, and if displayed on a 4:3 display (DAR = 4:3) has square pixels, hence a PAR of 1:1. By contrast, a 720 × 576 D-1 PAL image has a SAR of 720/576 = 5:4, but if displayed on a 4:3 display (DAR = 4:3) the PAR is 4/3 : 5/4 = 16:15 ≈ 1.066. This means that the pixels of the PAL picture must be "stretched" by this amount to fit in the 4:3 display. In analog images such as film there is no notion of pixel, nor notion of SAR or PAR, but in the digitization of analog images the resulting digital image has pixels, hence SAR (and accordingly PAR, if displayed at the same aspect ratio as the original). Non-square pixels arise often in early digital TV standards, related to digitalization of analog TV signals – whose vertical and "effective" horizontal resolutions differ and are thus best described by non-square pixels – and also in some digital video cameras and computer display modes, such as Color Graphics Adapter (CGA). Today they arise also in transcoding between resolutions with different SARs. Actual displays do not generally have non-square pixels, though digital sensors might; they are rather a mathematical abstraction used in resampling images to convert between resolutions. There are several complicating factors in understanding PAR, particularly as it pertains to digitization of analog video: First, analog video does not have pixels, but rather a raster scan, and thus has a well-defined vertical resolution (the lines of the raster), but not a well-defined horizontal resolution, since each line is an analog signal. However, by a standardized sampling rate, the effective horizontal resolution can be determined by the sampling theorem, as is done below. Second, due to overscan, some of the lines at the top and bottom of the raster are not visible, as are some of the possible image on the left and right – see Overscan: Analog to digital resolution issues. Also, the resolution may be rounded (DV NTSC uses 480 lines, rather than the 486 that are possible). Third, analog video signals are interlaced – each image (frame) is sent as two "fields", each with half the lines. Thus either the pixels are twice as tall as they would be without interlacing, or the image is deinterlaced. == Background == Video is presented as a sequential series of images called video frames. Historically, video frames were created and recorded in analog form. As digital display technology, digital broadcast technology, and digital video compression evolved separately, it resulted in video frame differences that must be addressed using pixel aspect ratio. Digital video frames are generally defined as a grid of pixels used to present each sequential image. The horizontal component is defined by pixels (or samples), and is known as a video line. The vertical component is defined by the number of lines, as in 480 lines. Standard-definition television standards and practices were developed as broadcast technologies and intended for terrestrial broadcasting, and were therefore not designed for digital video presentation. Such standards define an image as an array of well-defined horizontal "Lines", well-defined vertical "Line Duration" and a well-defined picture center. However, there is not a standard-definition television standard that properly defines image edges or explicitly demands a certain number of picture elements per line. Furthermore, analog video systems such as NTSC 480i and PAL 576i, instead of employing progressively displayed frames, employ fields or interlaced half-frames displayed in an interwoven manner to reduce flicker and double the image rate for smoother motion. === Analog-to-digital conversion === As a result of computers becoming powerful enough to serve as video editing tools, video digital-to-analog converters and analog-to-digital converters were made to overcome this incompatibility. To convert analog video lines into a series of square pixels, the industry adopted a default sampling rate at which luma values were extracted into pixels. The luma sampling rate for 480i pictures was 12+3⁄11 MHz and for 576i pictures was 14+3⁄4 MHz. The term pixel aspect ratio was first coined when ITU-R BT.601 (commonly known as Rec. 601) specified that standard-definition television pictures are made of lines of exactly 720 non-square pixels. ITU-R BT.601 did not define the exact pixel aspect ratio but did provide enough information to calculate the exact pixel aspect ratio based on industry practices: The standard luma sampling rate of precisely 13+1⁄2 MHz. Based on this information: The pixel aspect ratio for 480i would be 10:11 as: 12 3 11 ÷ 13 1 2 = 10 11 {\displaystyle 12{\tfrac {3}{11}}\div 13{\tfrac {1}{2}}={\tfrac {10}{11}}} The pixel aspect ratio for 576i would be 59:54 as: 14 3 4 ÷ 13 1 2 = 59 54 {\displaystyle 14{\tfrac {3}{4}}\div 13{\tfrac {1}{2}}={\tfrac {59}{54}}} SMPTE RP 187 further attempted to standardize the pixel aspect ratio values for 480i and 576i. It designated 177:160 for 480i or 1035:1132 for 576i. However, due to significant difference with practices in effect by industry and the computational load that they imposed upon the involved hardware, SMPTE RP 187 was simply ignored. SMPTE RP 187 information annex A.4 further suggested the use of 10:11 for 480i. As of this writing, ITU-R BT.601-6, which is the latest edition of ITU-R BT.601, still implies that the pixel aspect ratios mentioned above are correct. === Digital video processing === As stated above, ITU-R BT.601 specified that standard-definition television pictures are made of lines of 720 non-square pixels, sampled with a precisely specified sampling rate. A simple mathematical calculation reveals that a 704 pixel width would be enough to contain a 480i or 576i standard 4:3 picture: A 4:3 480-line picture, digitized with the Rec. 601-recommended sampling rate, would be 704 non-square pixels wide. x 480 × 10 11 = 4 3 ⇒ x = 480 × 11 × 4 10 × 3 = 704 {\displaystyle {\frac {x}{480}}\times {\frac {10}{11}}={\frac {4}{3}}\Rightarrow x={\frac {480\times 11\times 4}{10\times 3}}=704} A 4:3 576-line picture, digitized with the Rec. 601-recommended sampling rate, would be 702+54⁄59 non-square pixels wide. x 576 × 59 54 = 4 3 ⇒ x = 576 × 54 × 4 59 × 3 = 702 54 59 {\displaystyle {\frac {x}{576}}\times {\frac {59}{54}}={\frac {4}{3}}\Rightarrow x={\frac {576\times 54\times 4}{59\times 3}}=702{\tfrac {54}{59}}} Unfortunately, not all standard TV pictures are exactly 4:3: As mentioned earlier, in analog video, the center of a picture is well-defined but the edges of the picture are not standardized. As a result, some analog devices (mostly PAL devices but also some NTSC devices) generated motion pictures that were horizontally (slightly) wider. This also proportionately applies to anamorphic widescreen (16:9) pictures. Therefore, to maintain a safe margin of error, ITU-R BT.601 required sampling 16 more non-square pixels per line (8 more at each edge) to ensure saving all video data near the margins. This requirement, however, had implications for PAL motion pictures. PAL pixel aspect ratios for standard (4:3) and anamorphic wide screen (16:9), respectively 59:54 and 118:81, were awkward for digital image processing, especially for mixing PAL and NTSC video clips. Therefore, video editing products chose the almost equivalent value
Hybrid intelligent system
Hybrid intelligent system denotes a software system which employs, in parallel, a combination of methods and techniques from artificial intelligence subfields, such as: Neuro-symbolic systems Neuro-fuzzy systems Hybrid connectionist-symbolic models Fuzzy expert systems Connectionist expert systems Evolutionary neural networks Genetic fuzzy systems Rough fuzzy hybridization Reinforcement learning with fuzzy, neural, or evolutionary methods as well as symbolic reasoning methods. From the cognitive science perspective, every natural intelligent system is hybrid because it performs mental operations on both the symbolic and subsymbolic levels. For the past few years, there has been an increasing discussion of the importance of A.I. Systems Integration. Based on notions that there have already been created simple and specific AI systems (such as systems for computer vision, speech synthesis, etc., or software that employs some of the models mentioned above) and now is the time for integration to create broad AI systems. Proponents of this approach are researchers such as Marvin Minsky, Ron Sun, Aaron Sloman, Angelo Dalli and Michael A. Arbib. An example hybrid is a hierarchical control system in which the lowest, reactive layers are sub-symbolic. The higher layers, having relaxed time constraints, are capable of reasoning from an abstract world model and performing planning (even by hybrid wisdom). Intelligent systems usually rely on hybrid reasoning processes, which include induction, deduction, abduction and reasoning by analogy.
Deep Learning Super Sampling
Deep Learning Super Sampling (DLSS) is a suite of real-time deep learning image enhancement and upscaling technologies developed by Nvidia that are available in a number of video games. The goal of these technologies is to allow the majority of the graphics pipeline to run at a lower resolution for increased performance, and then infer a higher resolution image from this that approximates the same level of detail as if the image had been rendered at this higher resolution. This allows for higher graphical settings or frame rates for a given output resolution, depending on user preference. All generations of DLSS are available on all RTX-branded cards from Nvidia in supported titles. However, the Frame Generation feature is only supported on RTX 40 series GPUs or newer and Multi Frame Generation is only available on 50 series GPUs. == History == Nvidia advertised DLSS as a key feature of GeForce RTX 20 series GPUs when they launched in September 2018. At that time, the results were limited to a few video games, namely Battlefield V, or Metro Exodus, because the algorithm had to be trained specifically on each game on which it was applied and the results were usually not as good as simple resolution upscaling. In 2019, Control shipped with ray tracing and an image processing algorithm that approximated DLSS, which did not use the Tensor Cores. In April 2020, Nvidia advertised and shipped an improved version of DLSS named DLSS 2 with driver version 445.75. DLSS 2.0 was available for a few existing games including Control and Wolfenstein: Youngblood, and would later be added to many newly released games and game engines such as Unreal Engine and Unity. This time Nvidia said that it used the Tensor Cores again, and that the AI did not need to be trained specifically on each game. Despite sharing the DLSS branding, the two iterations of DLSS differ significantly and are not backwards-compatible. In January 2025, Nvidia stated that there are over 540 games and apps supporting DLSS, and that over 80% of Nvidia RTX users activate DLSS. In March 2025, there were more than 100 games that support DLSS 4, according to Nvidia. By May 2025, over 125 games supported DLSS 4. The first video game console to use DLSS, the Nintendo Switch 2, was released on June 5, 2025. Nvidia announced DLSS 4.5 at CES 2026. In January 2026, Nvidia stated that over 250 games and applications support Multi Frame Generation. On March 16, 2026, at GTC 2026, Nvidia CEO Jensen Huang presented DLSS 5, a real-time AI model based on neural rendering that realistically enhances lighting and material surfaces at up to 4K resolution while retaining the developer's intended art style. It is planned to release in fall of 2026. In a blog post on its website, Nvidia has announced that DLSS 5 will be available in such games as Assassin's Creed Shadows, Delta Force, Hogwarts Legacy, Naraka: Bladepoint, Phantom Blade Zero, Resident Evil Requiem, Starfield, The Elder Scrolls IV: Oblivion Remastered, and more. On May 31, 2026, Nvidia announced an updated version of Ray Reconstruction for DLSS 4.5 in a blog post, scheduled for release on all RTX GPUs in August of the same year. They said it is designed to better embed spatial awareness into scenes and analyze engine data on movements and lighting conditions, resulting in a sharper, more stable, and less noisy image. === Release timeline === == Technology == === DLSS 1 === The first iteration of DLSS is a predominantly spatial image upscaler with two stages, both relying on convolutional auto-encoder neural networks. The first step is an image enhancement network which uses the current frame and motion vectors to perform edge enhancement, and spatial anti-aliasing. The second stage is an image upscaling step which uses the single raw, low-resolution frame to upscale the image to the desired output resolution. Using just a single frame for upscaling means the neural network itself must generate a large amount of new information to produce the high-resolution output, which can result in slight hallucinations such as leaves that differ in style to the source content. The neural networks are trained on a per-game basis by generating a "perfect frame" using traditional supersampling to 64 samples per pixel, as well as the motion vectors for each frame. The data collected must be as comprehensive as possible, including as many levels, times of day, graphical settings, resolutions, etc. as possible. This data is also augmented using common augmentations such as rotations, colour changes, and random noise to help generalize the test data. Training is performed on Nvidia's Saturn V supercomputer. This first iteration received a mixed response, with many criticizing the often soft appearance and artifacts along with glitches in certain situations; likely a side effect of the limited data from only using a single frame input to the neural networks which could not be trained to perform optimally in all scenarios and edge-cases. Nvidia also demonstrated the ability for the auto-encoder networks to learn the ability to recreate depth-of-field and motion blur, although this functionality has never been included in a publicly released product. === DLSS 2 === DLSS 2 is a temporal anti-aliasing upsampling (TAAU) implementation, using data from previous frames extensively through sub-pixel jittering to resolve fine detail and reduce aliasing. The data DLSS 2 collects includes: the raw low-resolution input, motion vectors, depth buffers, and exposure / brightness information. It can also be used as a simpler TAA implementation where the image is rendered at 100% resolution, rather than being upsampled by DLSS, Nvidia brands this as DLAA (Deep Learning Anti-Aliasing). TAA(U) 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. DLSS 2 uses a convolutional auto-encoder neural network trained to identify and fix temporal artifacts, instead of manually programmed heuristics as mentioned above. Because of this, DLSS 2 can generally resolve detail better than other TAA and TAAU implementations, while also removing most temporal artifacts. This is why DLSS 2 can sometimes produce a sharper image than rendering at higher, or even native resolutions using traditional TAA. However, no temporal solution is perfect, and artifacts (ghosting in particular) are still visible in some scenarios when using DLSS 2. Because temporal artifacts occur in most art styles and environments in broadly the same way, the neural network that powers DLSS 2 does not need to be retrained when being used in different games. Despite this, Nvidia does frequently ship new minor revisions of DLSS 2 with new titles, so this could suggest some minor training optimizations may be performed as games are released, although Nvidia does not provide changelogs for these minor revisions to confirm this. The main advancements compared to DLSS 1 include: Significantly improved detail retention, a generalized neural network that does not need to be re-trained per-game, and ~2x less overhead (~1–2 ms vs ~2–4 ms). It should also be noted that forms of TAAU such as DLSS 2 are not upscalers in the same sense as techniques such as ESRGAN or DLSS 1, which attempt to create new information from a low-resolution source; instead, TAAU works to recover data from previous frames, rather than creating new data. In practice, this means low resolution textures in games will still appear low-resolution when using current TAAU techniques. This is why Nvidia recommends game developers use higher resolution textures than they would normally for a given rendering resolution by applying a mip-map bias when DLSS 2 is enabled. === DLSS 3 === Augments DLSS 2 with improved image quality and the introduction of a new motion interpolation feature, called Frame Generation. The DLSS Frame Generation algorithm takes two rendered frames from the rendering pipeline and generates a new frame that smoothly transitions between them. For every frame rendered, one additional frame is generated. DLSS 3.0 makes use of a new generation Optical Flow Accelerator (OFA) included in the Ada Lovelace architecture of GeForce RTX 40 series GPUs and with that is exclusive to them. The new OFA is said to be faster and more accurate than the one already available in previous Turing and Ampere RTX GPUs. === DLSS 3.5 === DLSS 3.5 adds Ray Reconstruction, replacing multiple denoising algorithms with a single AI model trained o
Admissible heuristic
In computer science, specifically in algorithms related to pathfinding, a heuristic function is said to be admissible if it never overestimates the cost of reaching the goal, i.e. the cost it estimates to reach the goal is not higher than the lowest possible cost from the current point in the path. In other words, it should act as a lower bound. It is related to the concept of consistent heuristics. While all consistent heuristics are admissible, not all admissible heuristics are consistent. == Search algorithms == An admissible heuristic is used to estimate the cost of reaching the goal state in an informed search algorithm. In order for a heuristic to be admissible to the search problem, the estimated cost must always be lower than or equal to the actual cost of reaching the goal state. The search algorithm uses the admissible heuristic to find an estimated optimal path to the goal state from the current node. For example, in A search the evaluation function (where n {\displaystyle n} is the current node) is: f ( n ) = g ( n ) + h ( n ) {\displaystyle f(n)=g(n)+h(n)} where f ( n ) {\displaystyle f(n)} = the evaluation function. g ( n ) {\displaystyle g(n)} = the cost from the start node to the current node h ( n ) {\displaystyle h(n)} = estimated cost from current node to goal. h ( n ) {\displaystyle h(n)} is calculated using the heuristic function. With a non-admissible heuristic, the A algorithm could overlook the optimal solution to a search problem due to an overestimation in f ( n ) {\displaystyle f(n)} . == Formulation == n {\displaystyle n} is a node h {\displaystyle h} is a heuristic h ( n ) {\displaystyle h(n)} is cost indicated by h {\displaystyle h} to reach a goal from n {\displaystyle n} h ∗ ( n ) {\displaystyle h^{}(n)} is the optimal cost to reach a goal from n {\displaystyle n} h ( n ) {\displaystyle h(n)} is admissible if, ∀ n {\displaystyle \forall n} h ( n ) ≤ h ∗ ( n ) {\displaystyle h(n)\leq h^{}(n)} == Construction == An admissible heuristic can be derived from a relaxed version of the problem, or by information from pattern databases that store exact solutions to subproblems of the problem, or by using inductive learning methods. == Examples == Two different examples of admissible heuristics apply to the fifteen puzzle problem: Hamming distance Manhattan distance The Hamming distance is the total number of misplaced tiles. It is clear that this heuristic is admissible since the total number of moves to order the tiles correctly is at least the number of misplaced tiles (each tile not in place must be moved at least once). The cost (number of moves) to the goal (an ordered puzzle) is at least the Hamming distance of the puzzle. The Manhattan distance of a puzzle is defined as: h ( n ) = ∑ all tiles d i s t a n c e ( tile, correct position ) {\displaystyle h(n)=\sum _{\text{all tiles}}{\mathit {distance}}({\text{tile, correct position}})} Consider the puzzle below in which the player wishes to move each tile such that the numbers are ordered. The Manhattan distance is an admissible heuristic in this case because every tile will have to be moved at least the number of spots in between itself and its correct position. The subscripts show the Manhattan distance for each tile. The total Manhattan distance for the shown puzzle is: h ( n ) = 3 + 1 + 0 + 1 + 2 + 3 + 3 + 4 + 3 + 2 + 4 + 4 + 4 + 1 + 1 = 36 {\displaystyle h(n)=3+1+0+1+2+3+3+4+3+2+4+4+4+1+1=36} == Optimality proof == If an admissible heuristic is used in an algorithm that, per iteration, progresses only the path of lowest evaluation (current cost + heuristic) of several candidate paths, terminates the moment its exploration reaches the goal and, crucially, closes all optimal paths before terminating (something that's possible with A search algorithm if special care isn't taken), then this algorithm can only terminate on an optimal path. To see why, consider the following proof by contradiction: Assume such an algorithm managed to terminate on a path T with a true cost Ttrue greater than the optimal path S with true cost Strue. This means that before terminating, the evaluated cost of T was less than or equal to the evaluated cost of S (or else S would have been picked). Denote these evaluated costs Teval and Seval respectively. The above can be summarized as follows, Strue < Ttrue Teval ≤ Seval If our heuristic is admissible it follows that at this penultimate step Teval = Ttrue because any increase on the true cost by the heuristic on T would be inadmissible and the heuristic cannot be negative. On the other hand, an admissible heuristic would require that Seval ≤ Strue which combined with the above inequalities gives us Teval < Ttrue and more specifically Teval ≠ Ttrue. As Teval and Ttrue cannot be both equal and unequal our assumption must have been false and so it must be impossible to terminate on a more costly than optimal path. As an example, let us say we have costs as follows:(the cost above/below a node is the heuristic, the cost at an edge is the actual cost) 0 10 0 100 0 START ---- O ----- GOAL | | 0| |100 | | O ------- O ------ O 100 1 100 1 100 So clearly we would start off visiting the top middle node, since the expected total cost, i.e. f ( n ) {\displaystyle f(n)} , is 10 + 0 = 10 {\displaystyle 10+0=10} . Then the goal would be a candidate, with f ( n ) {\displaystyle f(n)} equal to 10 + 100 + 0 = 110 {\displaystyle 10+100+0=110} . Then we would clearly pick the bottom nodes one after the other, followed by the updated goal, since they all have f ( n ) {\displaystyle f(n)} lower than the f ( n ) {\displaystyle f(n)} of the current goal, i.e. their f ( n ) {\displaystyle f(n)} is 100 , 101 , 102 , 102 {\displaystyle 100,101,102,102} . So even though the goal was a candidate, we could not pick it because there were still better paths out there. This way, an admissible heuristic can ensure optimality. However, note that although an admissible heuristic can guarantee final optimality, it is not necessarily efficient.
Seq2seq
Seq2seq is a family of machine learning approaches used for natural language processing. Originally developed by Lê Viết Quốc, a Vietnamese computer scientist and a machine learning pioneer at Google Brain, this framework has become foundational in many modern AI systems. Applications include language translation, image captioning, conversational models, speech recognition, and text summarization. Seq2seq uses sequence transformation: it turns one sequence into another sequence. == History == One naturally wonders if the problem of translation could conceivably be treated as a problem in cryptography. When I look at an article in Russian, I say: 'This is really written in English, but it has been coded in some strange symbols. I will now proceed to decode. seq2seq is an approach to machine translation (or more generally, sequence transduction) with roots in information theory, where communication is understood as an encode-transmit-decode process, and machine translation can be studied as a special case of communication. This viewpoint was elaborated, for example, in the noisy channel model of machine translation. In practice, seq2seq maps an input sequence into a real-numerical vector by using a neural network (the encoder), and then maps it back to an output sequence using another neural network (the decoder). The idea of encoder-decoder sequence transduction had been developed in the early 2010s. The papers most commonly cited as the originators that produced seq2seq are two papers from 2014. In the seq2seq as proposed by them, both the encoder and the decoder were LSTMs. This had the "bottleneck" problem, since the encoding vector has a fixed size, so for long input sequences, information would tend to be lost, as they are difficult to fit into the fixed-length encoding vector. The attention mechanism, proposed in 2014, resolved the bottleneck problem. They called their model RNNsearch, as it "emulates searching through a source sentence during decoding a translation". A problem with seq2seq models at this point was that recurrent neural networks are difficult to parallelize. The 2017 publication of Transformers resolved the problem by replacing the encoding RNN with self-attention Transformer blocks ("encoder blocks"), and the decoding RNN with cross-attention causally-masked Transformer blocks ("decoder blocks"). === Priority dispute === One of the papers cited as the originator for seq2seq is (Sutskever et al 2014), published at Google Brain while they were on Google's machine translation project. The research allowed Google to overhaul Google Translate into Google Neural Machine Translation in 2016. Tomáš Mikolov claims to have developed the idea (before joining Google Brain) of using a "neural language model on pairs of sentences... and then [generating] translation after seeing the first sentence"—which he equates with seq2seq machine translation, and to have mentioned the idea to Ilya Sutskever and Quoc Le (while at Google Brain), who failed to acknowledge him in their paper. Mikolov had worked on RNNLM (using RNN for language modelling) for his PhD thesis, and is more notable for developing word2vec. == Architecture == The main reference for this section is. === Encoder === The encoder is responsible for processing the input sequence and capturing its essential information, which is stored as the hidden state of the network and, in a model with attention mechanism, a context vector. The context vector is the weighted sum of the input hidden states and is generated for every time instance in the output sequences. === Decoder === The decoder takes the context vector and hidden states from the encoder and generates the final output sequence. The decoder operates in an autoregressive manner, producing one element of the output sequence at a time. At each step, it considers the previously generated elements, the context vector, and the input sequence information to make predictions for the next element in the output sequence. Specifically, in a model with attention mechanism, the context vector and the hidden state are concatenated together to form an attention hidden vector, which is used as an input for the decoder. The seq2seq method developed in the early 2010s uses two neural networks: an encoder network converts an input sentence into numerical vectors, and a decoder network converts those vectors to sentences in the target language. The Attention mechanism was grafted onto this structure in 2014 and is shown below. Later it was refined into the encoder-decoder Transformer architecture of 2017. === Training vs prediction === There is a subtle difference between training and prediction. During training time, both the input and the output sequences are known. During prediction time, only the input sequence is known, and the output sequence must be decoded by the network itself. Specifically, consider an input sequence x 1 : n {\displaystyle x_{1:n}} and output sequence y 1 : m {\displaystyle y_{1:m}} . The encoder would process the input x 1 : n {\displaystyle x_{1:n}} step by step. After that, the decoder would take the output from the encoder, as well as the
Empirical risk minimization
In statistical learning theory, the principle of empirical risk minimization defines a family of learning algorithms based on evaluating performance over a known and fixed dataset. The core idea is based on an application of the law of large numbers; more specifically, we cannot know exactly how well a predictive algorithm will work in practice (i.e. the "true risk") because we do not know the true distribution of the data, but we can instead estimate and optimize the performance of the algorithm on a known set of training data. The performance over the known set of training data is referred to as the "empirical risk". == Background == The following situation is a general setting of many supervised learning problems. There are two spaces of objects X {\displaystyle X} and Y {\displaystyle Y} and we would like to learn a function h : X → Y {\displaystyle \ h:X\to Y} (often called hypothesis) which outputs an object y ∈ Y {\displaystyle y\in Y} , given x ∈ X {\displaystyle x\in X} . To do so, there is a training set of n {\displaystyle n} examples ( x 1 , y 1 ) , … , ( x n , y n ) {\displaystyle \ (x_{1},y_{1}),\ldots ,(x_{n},y_{n})} where x i ∈ X {\displaystyle x_{i}\in X} is an input and y i ∈ Y {\displaystyle y_{i}\in Y} is the corresponding response that is desired from h ( x i ) {\displaystyle h(x_{i})} . To put it more formally, assuming that there is a joint probability distribution P ( x , y ) {\displaystyle P(x,y)} over X {\displaystyle X} and Y {\displaystyle Y} , and that the training set consists of n {\displaystyle n} instances ( x 1 , y 1 ) , … , ( x n , y n ) {\displaystyle \ (x_{1},y_{1}),\ldots ,(x_{n},y_{n})} drawn i.i.d. from P ( x , y ) {\displaystyle P(x,y)} . The assumption of a joint probability distribution allows for the modelling of uncertainty in predictions (e.g. from noise in data) because y {\displaystyle y} is not a deterministic function of x {\displaystyle x} , but rather a random variable with conditional distribution P ( y | x ) {\displaystyle P(y|x)} for a fixed x {\displaystyle x} . It is also assumed that there is a non-negative real-valued loss function L ( y ^ , y ) {\displaystyle L({\hat {y}},y)} which measures how different the prediction y ^ {\displaystyle {\hat {y}}} of a hypothesis is from the true outcome y {\displaystyle y} . For classification tasks, these loss functions can be scoring rules. The risk associated with hypothesis h ( x ) {\displaystyle h(x)} is then defined as the expectation of the loss function: R ( h ) = E [ L ( h ( x ) , y ) ] = ∫ L ( h ( x ) , y ) d P ( x , y ) . {\displaystyle R(h)=\mathbf {E} [L(h(x),y)]=\int L(h(x),y)\,dP(x,y).} A loss function commonly used in theory is the 0-1 loss function: L ( y ^ , y ) = { 1 if y ^ ≠ y 0 if y ^ = y {\displaystyle L({\hat {y}},y)={\begin{cases}1&{\mbox{ if }}\quad {\hat {y}}\neq y\\0&{\mbox{ if }}\quad {\hat {y}}=y\end{cases}}} . The ultimate goal of a learning algorithm is to find a hypothesis h ∗ {\displaystyle h^{}} among a fixed class of functions H {\displaystyle {\mathcal {H}}} for which the risk R ( h ) {\displaystyle R(h)} is minimal: h ∗ = a r g m i n h ∈ H R ( h ) . {\displaystyle h^{}={\underset {h\in {\mathcal {H}}}{\operatorname {arg\,min} }}\,{R(h)}.} For classification problems, the Bayes classifier is defined to be the classifier minimizing the risk defined with the 0–1 loss function. == Formal definition == In general, the risk R ( h ) {\displaystyle R(h)} cannot be computed because the distribution P ( x , y ) {\displaystyle P(x,y)} is unknown to the learning algorithm. However, given a sample of iid training data points, we can compute an estimate, called the empirical risk, by computing the average of the loss function over the training set; more formally, computing the expectation with respect to the empirical measure: R emp ( h ) = 1 n ∑ i = 1 n L ( h ( x i ) , y i ) . {\displaystyle \!R_{\text{emp}}(h)={\frac {1}{n}}\sum _{i=1}^{n}L(h(x_{i}),y_{i}).} The empirical risk minimization principle states that the learning algorithm should choose a hypothesis h ^ {\displaystyle {\hat {h}}} which minimizes the empirical risk over the hypothesis class H {\displaystyle {\mathcal {H}}} : h ^ = a r g m i n h ∈ H R emp ( h ) . {\displaystyle {\hat {h}}={\underset {h\in {\mathcal {H}}}{\operatorname {arg\,min} }}\,R_{\text{emp}}(h).} Thus, the learning algorithm defined by the empirical risk minimization principle consists in solving the above optimization problem. == Properties == Guarantees for the performance of empirical risk minimization depend strongly on the function class selected as well as the distributional assumptions made. In general, distribution-free methods are too coarse, and do not lead to practical bounds. However, they are still useful in deriving asymptotic properties of learning algorithms, such as consistency. In particular, distribution-free bounds on the performance of empirical risk minimization given a fixed function class can be derived using bounds on the VC complexity of the function class. For simplicity, considering the case of binary classification tasks, it is possible to bound the probability of the selected classifier, ϕ n {\displaystyle \phi _{n}} being much worse than the best possible classifier ϕ ∗ {\displaystyle \phi ^{}} . Consider the risk L {\displaystyle L} defined over the hypothesis class C {\displaystyle {\mathcal {C}}} with growth function S ( C , n ) {\displaystyle {\mathcal {S}}({\mathcal {C}},n)} given a dataset of size n {\displaystyle n} . Then, for every ϵ > 0 {\displaystyle \epsilon >0} : P ( L ( ϕ n ) − L ( ϕ ∗ ) > ϵ ) ≤ 8 S ( C , n ) exp { − n ϵ 2 / 32 } {\displaystyle \mathbb {P} \left(L(\phi _{n})-L(\phi ^{})>\epsilon \right)\leq {\mathcal {8}}S({\mathcal {C}},n)\exp\{-n\epsilon ^{2}/32\}} Similar results hold for regression tasks. These results are often based on uniform laws of large numbers, which control the deviation of the empirical risk from the true risk, uniformly over the hypothesis class. === Impossibility results === It is also possible to show lower bounds on algorithm performance if no distributional assumptions are made. This is sometimes referred to as the No free lunch theorem. Even though a specific learning algorithm may provide the asymptotically optimal performance for any distribution, the finite sample performance is always poor for at least one data distribution. This means that no classifier can improve on the error for a given sample size for all distributions. Specifically, let ϵ > 0 {\displaystyle \epsilon >0} and consider a sample size n {\displaystyle n} and classification rule ϕ n {\displaystyle \phi _{n}} , there exists a distribution of ( X , Y ) {\displaystyle (X,Y)} with risk L ∗ = 0 {\displaystyle L^{}=0} (meaning that perfect prediction is possible) such that: E L n ≥ 1 / 2 − ϵ . {\displaystyle \mathbb {E} L_{n}\geq 1/2-\epsilon .} It is further possible to show that the convergence rate of a learning algorithm is poor for some distributions. Specifically, given a sequence of decreasing positive numbers a i {\displaystyle a_{i}} converging to zero, it is possible to find a distribution such that: E L n ≥ a i {\displaystyle \mathbb {E} L_{n}\geq a_{i}} for all n {\displaystyle n} . This result shows that universally good classification rules do not exist, in the sense that the rule must be low quality for at least one distribution. === Computational complexity === Empirical risk minimization for a classification problem with a 0-1 loss function is known to be an NP-hard problem even for a relatively simple class of functions such as linear classifiers. Nevertheless, it can be solved efficiently when the minimal empirical risk is zero, i.e., data is linearly separable. In practice, machine learning algorithms cope with this issue either by employing a convex approximation to the 0–1 loss function (like hinge loss for SVM), which is easier to optimize, or by imposing assumptions on the distribution P ( x , y ) {\displaystyle P(x,y)} (and thus stop being agnostic learning algorithms to which the above result applies). In the case of convexification, Zhang's lemma majors the excess risk of the original problem using the excess risk of the convexified problem. Minimizing the latter using convex optimization also allow to control the former. == Tilted empirical risk minimization == Tilted empirical risk minimization is a machine learning technique used to modify standard loss functions like squared error, by introducing a tilt parameter. This parameter dynamically adjusts the weight of data points during training, allowing the algorithm to focus on specific regions or characteristics of the data distribution. Tilted empirical risk minimization is particularly useful in scenarios with imbalanced data or when there is a need to emphasize errors in certain parts of the prediction space.