AI Detector Generator

AI Detector Generator — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

Nouvelle AI

Nouvelle artificial intelligence (Nouvelle AI) is an approach to artificial intelligence pioneered in the 1980s by Rodney Brooks, who was then part of MIT artificial intelligence laboratory. Nouvelle AI differs from classical AI by aiming to produce robots with intelligence levels similar to insects. Researchers believe that intelligence can emerge organically from simple behaviors as these intelligences interacted with the "real world", instead of using the constructed worlds which symbolic AIs typically needed to have programmed into them. == Motivation == The differences between nouvelle AI and symbolic AI are apparent in early robots Shakey and Freddy. These robots contained an internal model (or "representation") of their micro-worlds consisting of symbolic descriptions. As a result, this structure of symbols had to be renewed as the robot moved or the world changed. Shakey's planning programs assessed the program structure and broke it down into the necessary steps to complete the desired action. This level of computation required a large amount time to process, so Shakey typically performed its tasks very slowly. Symbolic AI researchers had long been plagued by the problem of updating, searching, and otherwise manipulating the symbolic worlds inside their AIs. A nouvelle system refers continuously to its sensors rather than to an internal model of the world. It processes the external world information it needs from the senses when it is required. As Brooks puts it, "the world is its own best model--always exactly up to date and complete in every detail." A central idea of nouvelle AI is that simple behaviors combine to form more complex behaviors over time. For example, simple behaviors can include elements like "move forward" and "avoid obstacles." A robot using nouvelle AI with simple behaviors like collision avoidance and moving toward a moving object could possibly come together to produce a more complex behavior like chasing a moving object. === The frame problem === The frame problem describes an issue with using first-order logic (FOL) to express facts about a robot in the world. Representing the state of a robot with traditional FOL requires the use of many axioms (symbolic language) to imply that things about an environment do not change arbitrarily. Nouvelle AI seeks to sidestep the frame problem by dispensing with filling the AI or robot with volumes of symbolic language and instead letting more complex behaviors emerge by combining simpler behavioral elements. === Embodiment === The goal of traditional AI was to build intelligences without bodies, which would only have been able to interact with the world via keyboard, screen, or printer. However, nouvelle AI attempts to build embodied intelligence situated in the real world. Brooks quotes approvingly from the brief sketches that Turing gave in 1948 and 1950 of the "situated" approach. Turing wrote of equipping a machine "with the best sense organs that money can buy" and teaching it "to understand and speak English" by a process that would "follow the normal teaching of a child." This approach was contrasted to the others where they focused on abstract activities such as playing chess. == Brooks' robots == === Insectoid robots === Brooks focused on building robots that acted like simple insects while simultaneously working to remove some traditional AI characteristics. He created insect-like robots, named Allen and Herbert after cognitive science and AI pioneers Allen Newell and Herbert A. Simon. Brooks's insectoid robots contained no internal models of the world. Herbert, for example, discarded a high volume of the information received from its sensors and never stored information for more than two seconds. ==== Allen ==== Allen had a ring of twelve ultrasonic sonars as its primary sensors and three independent behavior-producing modules. These modules were programmed to avoid both stationary and moving objects. With only this module activated, Allen stayed in the middle of a room until an object approached and then it ran away while avoiding obstacles in its way. ==== Herbert ==== Herbert used infrared sensors to avoid obstacles and a laser system to collect 3D data over a distance of about 12 feet. Herbert also carried a number of simple sensors in its "hand." The robot's testing ground was the real world environment of the busy offices and workspaces of the MIT AI lab where it searched for empty soda cans and carried them away, a seemingly goal-oriented activity that emerged as a result of 15 simple behavior units combining. As a parallel, Simon noted that an ant's complicated path is due to the structure of its environment rather than the depth of its thought processes. ==== Other insectoid robots ==== Other robots by Brooks' team were Genghis and Squirt. Genghis had six legs and was able to walk over rough terrain and follow a human. Squirt's behavior modules had it stay in dark corners until it heard a noise, then it would begin to follow the source of the noise. Brooks agreed that the level of nouvelle AI had come near the complexity of a real insect, which raised a question about whether or not insect level-behavior was and is a reasonable goal for nouvelle AI. === Humanoid robots === Brooks' own recent work has taken the opposite direction to that proposed by Von Neumann in the quotations "theorists who select the human nervous system as their model are unrealistically picking 'the most complicated object under the sun,' and that there is little advantage in selecting instead the ant, since any nervous system at all exhibits exceptional complexity." ==== Cog ==== In the 1990s, Brooks decided to pursue the goal of human-level intelligence and, with Lynn Andrea Stein, built a humanoid robot called Cog. Cog is a robot with an extensive collection of sensors, a face, and arms (among other features) that allow it to interact with the world and gather information and experience so as to assemble intelligence organically in the manner described above by Turing. The team believed that Cog would be able to learn and able to find a correlation between the sensory information it received and its actions, and to learn common sense knowledge on its own. As of 2003, all development of the project had ceased.
Read more →
Comparison of raster graphics editors

Raster graphics editors can be compared by many variables, including availability. == List == == General information == Basic general information about the editor: creator, company, license, etc. == Operating system support == The operating systems on which the editors can run natively, that is, without emulation, virtual machines or compatibility layers. In other words, the software must be specifically coded for the operation system; for example, Adobe Photoshop for Windows running on Linux with Wine does not fit. == Features == == Color spaces == == File support ==
Read more →
Kernel (image processing)

In image processing, a kernel, convolution matrix, or mask is a small matrix used for blurring, sharpening, embossing, edge detection, and more. This is accomplished by doing a convolution between the kernel and an image. Or more simply, when each pixel in the output image is a function of the nearby pixels (including itself) in the input image, the kernel is that function. == Details == The general expression of a convolution is g x , y = ω ∗ f x , y = ∑ i = − a a ∑ j = − b b ω i , j f x − i , y − j , {\displaystyle g_{x,y}=\omega f_{x,y}=\sum _{i=-a}^{a}{\sum _{j=-b}^{b}{\omega _{i,j}f_{x-i,y-j}}},} where g ( x , y ) {\displaystyle g(x,y)} is the filtered image, f ( x , y ) {\displaystyle f(x,y)} is the original image, ω {\displaystyle \omega } is the filter kernel. Every element of the filter kernel is considered by − a ≤ i ≤ a {\displaystyle -a\leq i\leq a} and − b ≤ j ≤ b {\displaystyle -b\leq j\leq b} . Depending on the element values, a kernel can cause a wide range of effects: The above are just a few examples of effects achievable by convolving kernels and images. === Origin === The origin is the position of the kernel which is above (conceptually) the current output pixel. This could be outside of the actual kernel, though usually it corresponds to one of the kernel elements. For a symmetric kernel, the origin is usually the center element. == Convolution == Convolution is the process of adding each element of the image to its local neighbors, weighted by the kernel. This is related to a form of mathematical convolution. The matrix operation being performed—convolution—is not traditional matrix multiplication, despite being similarly denoted by . For example, if we have two three-by-three matrices, the first a kernel, and the second an image piece, convolution is the process of flipping both the rows and columns of the kernel and multiplying locally similar entries and summing. The element at coordinates [2, 2] (that is, the central element) of the resulting image would be a weighted combination of all the entries of the image matrix, with weights given by the kernel: ( [ a b c d e f g h i ] ∗ [ 1 2 3 4 5 6 7 8 9 ] ) [ 2 , 2 ] = {\displaystyle \left({\begin{bmatrix}a&b&c\\d&e&f\\g&h&i\end{bmatrix}}{\begin{bmatrix}1&2&3\\4&5&6\\7&8&9\end{bmatrix}}\right)[2,2]=} ( i ⋅ 1 ) + ( h ⋅ 2 ) + ( g ⋅ 3 ) + ( f ⋅ 4 ) + ( e ⋅ 5 ) + ( d ⋅ 6 ) + ( c ⋅ 7 ) + ( b ⋅ 8 ) + ( a ⋅ 9 ) . {\displaystyle (i\cdot 1)+(h\cdot 2)+(g\cdot 3)+(f\cdot 4)+(e\cdot 5)+(d\cdot 6)+(c\cdot 7)+(b\cdot 8)+(a\cdot 9).} The other entries would be similarly weighted, where we position the center of the kernel on each of the boundary points of the image, and compute a weighted sum. The values of a given pixel in the output image are calculated by multiplying each kernel value by the corresponding input image pixel values. This can be described algorithmically with the following pseudo-code: for each image row in input image: for each pixel in image row: set accumulator to zero for each kernel row in kernel: for each element in kernel row: if element position corresponding to pixel position then multiply element value corresponding to pixel value add result to accumulator endif set output image pixel to accumulator corresponding input image pixels are found relative to the kernel's origin. If the kernel is symmetric then place the center (origin) of the kernel on the current pixel. The kernel will overlap the neighboring pixels around the origin. Each kernel element should be multiplied with the pixel value it overlaps with and all of the obtained values should be summed. This resultant sum will be the new value for the current pixel currently overlapped with the center of the kernel. If the kernel is not symmetric, it has to be flipped both around its horizontal and vertical axis before calculating the convolution as above. The general form for matrix convolution is [ x 11 x 12 ⋯ x 1 n x 21 x 22 ⋯ x 2 n ⋮ ⋮ ⋱ ⋮ x m 1 x m 2 ⋯ x m n ] ∗ [ y 11 y 12 ⋯ y 1 n y 21 y 22 ⋯ y 2 n ⋮ ⋮ ⋱ ⋮ y m 1 y m 2 ⋯ y m n ] = ∑ i = 0 m − 1 ∑ j = 0 n − 1 x ( m − i ) ( n − j ) y ( 1 + i ) ( 1 + j ) {\displaystyle {\begin{bmatrix}x_{11}&x_{12}&\cdots &x_{1n}\\x_{21}&x_{22}&\cdots &x_{2n}\\\vdots &\vdots &\ddots &\vdots \\x_{m1}&x_{m2}&\cdots &x_{mn}\\\end{bmatrix}}{\begin{bmatrix}y_{11}&y_{12}&\cdots &y_{1n}\\y_{21}&y_{22}&\cdots &y_{2n}\\\vdots &\vdots &\ddots &\vdots \\y_{m1}&y_{m2}&\cdots &y_{mn}\\\end{bmatrix}}=\sum _{i=0}^{m-1}\sum _{j=0}^{n-1}x_{(m-i)(n-j)}y_{(1+i)(1+j)}} === Edge handling === Kernel convolution usually requires values from pixels outside of the image boundaries. There are a variety of methods for handling image edges. Extend The nearest border pixels are conceptually extended as far as necessary to provide values for the convolution. Corner pixels are extended in 90° wedges. Other edge pixels are extended in lines. Wrap The image is conceptually wrapped (or tiled) and values are taken from the opposite edge or corner. Mirror The image is conceptually mirrored at the edges. For example, attempting to read a pixel 3 units outside an edge reads one 3 units inside the edge instead. Crop / Avoid overlap Any pixel in the output image which would require values from beyond the edge is skipped. This method can result in the output image being slightly smaller, with the edges having been cropped. Move kernel so that values from outside of image is never required. Machine learning mainly uses this approach. Example: Kernel size 10x10, image size 32x32, result image is 23x23. Kernel Crop Any pixel in the kernel that extends past the input image isn't used and the normalizing is adjusted to compensate. Constant Use constant value for pixels outside of image. Usually black or sometimes gray is used. Generally this depends on application. === Normalization === Normalization is defined as the division of each element in the kernel by the sum of all kernel elements, so that the sum of the elements of a normalized kernel is unity. This will ensure the average pixel in the modified image is as bright as the average pixel in the original image. === Optimization === Fast convolution algorithms include: separable convolution ==== Separable convolution ==== 2D convolution with an M × N kernel requires M × N multiplications for each sample (pixel). If the kernel is separable, then the computation can be reduced to M + N multiplications. Using separable convolutions can significantly decrease the computation by doing 1D convolution twice instead of one 2D convolution. === Implementation === Here a concrete convolution implementation done with the GLSL shading language :
Read more →
Line detection

In image processing, line detection is an algorithm that takes a collection of n edge points and finds all the lines on which these edge points lie. The most popular line detectors are the Hough transform and convolution-based techniques. == Hough transform == The Hough transform can be used to detect lines and the output is a parametric description of the lines in an image, for example ρ = r cos(θ) + c sin(θ). If there is a line in a row and column based image space, it can be defined ρ, the distance from the origin to the line along a perpendicular to the line, and θ, the angle of the perpendicular projection from the origin to the line measured in degrees clockwise from the positive row axis. Therefore, a line in the image corresponds to a point in the Hough space. The Hough space for lines has therefore these two dimensions θ and ρ, and a line is represented by a single point corresponding to a unique set of these parameters. The Hough transform can then be implemented by choosing a set of values of ρ and θ to use. For each pixel (r, c) in the image, compute r cos(θ) + c sin(θ) for each values of θ, and place the result in the appropriate position in the (ρ, θ) array. At the end, the values of (ρ, θ) with the highest values in the array will correspond to strongest lines in the image == Convolution-based technique == In a convolution-based technique, the line detector operator consists of a convolution masks tuned to detect the presence of lines of a particular width n and a θ orientation. Here are the four convolution masks to detect horizontal, vertical, oblique (+45 degrees), and oblique (−45 degrees) lines in an image. a) Horizontal mask(R1) (b) Vertical (R3) (C) Oblique (+45 degrees)(R2) (d) Oblique (−45 degrees)(R4) In practice, masks are run over the image and the responses are combined given by the following equation: R(x, y) = max(|R1 (x, y)|, |R2 (x, y)|, |R3 (x, y)|, |R4 (x, y)|) If R(x, y) > T, then discontinuity As can be seen below, if mask is overlay on the image (horizontal line), multiply the coincident values, and sum all these results, the output will be the (convolved image). For example, (−1)(0)+(−1)(0)+(−1)(0) + (2)(1) +(2)(1)+(2)(1) + (−1)(0)+(−1)(0)+(−1)(0) = 6 pixels on the second row, second column in the (convolved image) starting from the upper left corner of the horizontal lines. page 82 == Example == These masks above are tuned for light lines against a dark background, and would give a big negative response to dark lines against a light background. == Code example == The code was used to detect only the vertical lines in an image using Matlab and the result is below. The original image is the one on the top and the result is below it. As can be seen on the picture on the right, only the vertical lines were detected
Read more →
StatMuse

StatMuse Inc. is an American artificial intelligence company founded in 2014. It operates an eponymous website that hosts a database of sports statistics covering the four major North American sports leagues, the Women's National Basketball Association (WNBA), NCAA Division I men's basketball, NCAA Division I Football Bowl Subdivision, the Big Five association football leagues in Europe, and various professional golf tours. == History == The company was founded by friends Adam Elmore and Eli Dawson in 2014. In email correspondence to the Springfield News-Leader, Elmore detailed that he and Dawson, fans of the National Basketball Association (NBA), were compelled to create StatMuse after they realized there was no online platform where they could search "Lebron James most points" [sic] and quickly get a result "showing his highest scoring games." As a startup, the company's goal was to utilize a type of artificial intelligence called natural language processing (NLP) for sports. In 2015, the company was part of the second group of startups accepted into the Disney Accelerator program. The company secured support from several investors, including The Walt Disney Company, Techstars, Allen & Company, the NFL Players Association, Greycroft and NBA Commissioner David Stern. As part of their partnership with Disney, StatMuse signed a content deal with ESPN (owned by Disney) to provide stats content on social media and television during the 2015–16 NBA season. Initially, the company only had stats available for the NBA, but eventually expanded to provide stats for the other major North American sports leagues. The company's initial demographic was players of fantasy sports, but it eventually expanded to target general sports fans as well. StatMuse offers responses to user queries in the voices of sports-related public figures. Dawson shared with VentureBeat that StatMuse brings people in and records them saying different words and phrases. These celebrity voices were made accessible through Google's Google Assistant service, Microsoft's Cortana virtual assistant, and Amazon's Echo devices. The company launched its phone app in September 2017. The app allows users to access StatMuse's sports statistics database by submitting queries in their natural language. Upon the launch of the phone app, Fitz Tepper of TechCrunch wrote that: "The technology isn't perfect – some of the pauses between words are a bit awkward, making it clear that some phrases are being stitched together on the fly. But this is the exception, and on the whole, most responses sound pretty good." StatMuse plug-ins for Slack and Facebook Messenger were also made, providing text-based sports stats. In 2019, StatMuse received investment from the Google Assistant Investment program. The service launched a premium option dubbed StatMuse+ in May 2023, offering options that had previously been included for free, such as unlimited searches and full results in data tables. The premium version also included early access to new features and a personalized search history, as well as not having ads. The app received a variety of feedback. In January 2024, the service launched a Premier League version of the website dubbed StatMuse FC. It is planned to introduce more leagues on the website.
Read more →
Alias Eclipse

Eclipse was a professional 2D image editing program available on Silicon Graphics and Windows workstations. Designed to manipulate high-resolution images like digitized movie frames and photographs for print, it offered color correction tools, image processing effects, rudimentary paint features, and spline-based drawing and masking. == History == Eclipse was originally developed in the late 1980s by Full Color Computing, an early provider of photo retouch and color prepress software for Silicon Graphics workstations. Alias Research (later Alias Systems Corporation), a developer of professional 3D graphics applications for the SGI platform, purchased the rights to Eclipse in fall 1990. Alias developed Eclipse through the early to mid-1990s, releasing version 2.5 in 1995 with improvements to the speed of color correction, effects, and rendering. Xyvision's Contex Prepress division purchased exclusive rights to Eclipse from Alias in 1996, and released version 3.0 the following year. Eclipse was subsequently sold to German developer Form & Vision GmbH, which continued development and ported it to the Windows platform. In 1999, Form & Vision released a demo of Eclipse 3.1.3 on the SGI platform which was limited to 1600 x 1600 pixel images, then ceased development of Eclipse on the SGI platform. Eclipse was thereafter developed exclusively for the Windows platform, culminating with version 3.1.4 in 2001. In the same year the firm went bankrupt. == Features == Eclipse was designed to work with very large images that could not be manipulated in real time on contemporary computer systems due to memory limitations, and thus allowed the user to make modifications to a lower-resolution copy of the original image in "proxy mode." Brush strokes, color corrections, and other edits were saved in proxy mode, then applied to the full-size image in post processing. This method also allowed for batch processing of a high-resolution image sequence using the edits applied to the original proxy image. Other features included color correction and separation, warping, special effects, text, and shape masking. Wavelet image compression created by LuraTech was added to Eclipse 3.1.4
Read more →
LCD crosstalk

LCD crosstalk is a visual defect in an LCD screen which occurs because of interference between adjacent pixels. Owing to the way rows and columns in the display are addressed, and charge is pushed around, the data on one part of the display has the potential to influence what is displayed elsewhere. This is generally known as crosstalk, and in matrix displays typically occurs in the horizontal and vertical directions. Crosstalk used to be a serious problem in the old passive-matrix (STN) displays, but is rarely discernable in modern active-matrix (TFT) displays. A fortunate side effect of inversion (see above) is that, for most display material, what little crosstalk there is largely cancelled out. For most practical purposes, the level of crosstalk in modern LCDs is negligible. Certain patterns, particularly those involving fine dots, can interact with the inversion and reveal visible crosstalk. If you try moving a small Window in front of the inversion pattern (above) which makes your screen flicker the most, you may well see crosstalk in the surrounding pattern. Different patterns are required to reveal crosstalk on different displays (depending on their inversion scheme).
Read more →
Pixel shift

Pixel shift is a method in digital cameras for producing a super-resolution image. The method works by taking several images, after each such capture moving ("shifting") the sensor to a new position. In digital colour cameras that employ pixel shift, this avoids a major limitation inherent in using Bayer pattern for obtaining colour, and instead produces an image with increased colour resolution and, assuming a static subject or additional computational steps, an image free of colour moiré. Taking this idea further, sub-pixel shifting may increase the resolution of the final image beyond that suggested by the specified resolution of the image sensor. Additionally, assuming that the various individual captures are taken at the same sensitivity, the final combined image will have less image noise than a single capture. This can be thought of as an averaging effect (for instance, in a pixel shift image composed of four individual frames with a classic Bayer pattern, every pixel in the final colour image is based on two measurements of the green channel). == List of cameras implementing pixel shift == All of the following cameras are fabricated with one imaging sensor, thus any kind of pixel shift requires a movement of the whole sensor. === Canon === Canon R5: Contains a 45 Mpixel sensor. The High-Resolution Mode shifts the sensor by one pixel to obtain a sequence of nine images that are merged into a 400 Mpixel image. === Fujifilm === Fujifilm GFX50S II: contains a 51 Mpixel sensor. The Pixel Shift Multi-Shot mode shifts the imaging sensor by 0.5-pixel movements to obtain a sequence of 16 images that are subsequently merged into a 200 Mpixel image. Fujifilm GFX100, Fujifilm GFX100 II: contains a 102 Mpixel sensor. A sequence of 16 pixel shifted images are merged into a 400 Mpixel image. Fujifilm GFX100S, Fujifilm GFX100S II: contains a 102 Mpixel sensor. A sequence of 16 pixel shifted images are merged into a 400 Mpixel image Fujifilm GFX100IR: contains a 102 Mpixel sensor. A sequence of 16 pixel shifted images are merged into a 400 Mpixel image Fujifilm X-H2: contains a 40 Mpixel sensor. A sequence of 20 shifted images are merged into a 160 Mpixel image. Fujifilm X-T5: contains a 40 Mpixel sensor. A sequence of 20 shifted images are merged into a 160 Mpixel image. === Nikon === Nikon Z8: contains a 47.5 Mpixel sensor. The High Res shot mode shifts the imaging sensor by 0.5-pixel movements to obtain a sequence of up to 32 images that can be merged in Nikon's NX studio software. Nikon Zf: contains a 24 Mpixel sensor. The High Res shot mode shifts the imaging sensor by 0.5-pixel movements to obtain a sequence of up to 32 images that can be merged in Nikon's NX studio software. === Olympus === Olympus OM-D E-M1 Mark II: contains a 20.4 Mpixel sensor. The High Res shot mode produces a 50 Mpixel image. Olympus OM-D E-M5 Mark II: contains a 16 Mpixel sensor. The High Res shot mode shifts the imaging sensor by 0.5-pixel movements to obtain a sequence of 8 images that are subsequently merged into a 40 Mpixel image. Olympus OM-D E-M5 Mark III: contains a 20.4 Mpixel sensor. The High Res shot mode shifts the imaging sensor by 0.5-pixel movements to obtain a sequence of 8 images that are subsequently merged into a 50 Mpixel image. Olympus OM-D E-M1X: contains a 20.4 Mpixel sensor. The camera sports two pixel shift mode: (a) the 80Mp Tripod mode produces an 80 Mpixel image, (b) the Handheld High Res shot mode produces a 50 Mpixel image. Olympus PEN-F: contains a 20.4 Mpixel sensor. The High Res Shot mode takes multiple images, continually shifting the position of the sensor in sub-pixel increments. Combining these images results in either a 50MP JPEG or an 80MP Raw file. ==== OM System ==== OM System OM-1: contains a 20MPix sensor. The High Res Shot mode takes multiple images, and it can be used handheld or on a tripod. Handheld it will internally produce 50 Mpix files and 80 Mpix when mounted on a tripod. OM System OM-5: contains a 20MPix sensor. The High Res Shot mode takes multiple images, and it can be used handheld or on a tripod. Handheld it will internally produce 50 Mpix files and 80 Mpix when mounted on a tripod. === Panasonic === Panasonic Lumix DC-G9: contains a 20.3 Mpixel sensor. The High Resolution Mode takes a sequence of 8 shots in quick succession between which the sensor is shifted by 0.5 pixel for each image. These are subsequently merged into an 80 Mpixel image. Panasonic Lumix DC-S1: contains a 24.2 Mpixel sensor. The High Resolution Mode takes a sequence of shots in quick succession between which the sensor is shifted by a small amount. These are subsequently merged into a 96 Mpixel image. Panasonic Lumix DC-S1R: contains a 47.3 Mpixel sensor. The High Resolution Mode shifts the imaging sensor by a small increments to obtain a sequence of 8 images that are subsequently merged into a 187 Mpixel image. Panasonic Lumix DC-S1H Panasonic Lumix DC-S5 === Pentax === Pentax K-70: contains a 24.3 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'all color data in each pixel to deliver super-high-resolution images'. Pentax KP: contains a 24.3 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'high-resolution images with more accurate colours and much finer details'. Pentax K-3 II: contains a 24.3 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'super-high-resolution images with far more truthful color reproduction and much finer details'. Pentax K-3 III: contains a 25.7 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'a cancelling out of the Bayer pattern and removal of the need for sharpness-sapping demosaicing'. Pentax K-1: contains a 36.4 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'improved detail and colour resolution'. Pentax K-1 II: contains a 36.4 Mpixel sensor. The camera sports two pixel shift mode: (a) a series of 4 tripod-stabilised images shifted by 1 pixel each are subsequently combined into a 47.3 Mpixel image, (b) a series of images taken in handheld mode are combined into a 47.3 Mpixel image that is, within limits, able to cope even with moving subjects. === Sony === Sony a6600: contains a 24.3 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into an image sporting 'all color data in each pixel to deliver super-high-resolution images'. Sony α7R III: contains a 42.4 Mpixel sensor. The pixel shift mode takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into a 42.4 Mpixel image with improved tonal resolution. Sony α7R IV: contains a 61 Mpixel sensor. The camera has two pixel shift modes, (a) the first takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into a 61 Mpixel image with improved tonal resolution, (b) the other takes a sequence of 16 shots between which the sensor is shifted by 0.5 pixel. These are subsequently merged into a 240 Mpixel image with both enhanced detail and improved tonal resolution. Sony α1: contains a 50 Mpixel sensor. The camera has two pixel shift modes, (a) the first takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into a 50 Mpixel image with improved tonal resolution, (b) the other takes a sequence of 16 shots between which the sensor is shifted by 0.5 pixel. These are subsequently merged into a 200 Mpixel image with both enhanced detail and improved tonal resolution. === Hasselblad === Hasselblad H3DII: the model H3DII-39 sports a 39 Mpixel sensor, the model H3DII-50 a 50 Mpixel sensor. Both enable a pixel shift mode which takes a sequence of 4 shots between which the sensor is shifted by 1 pixel. These are subsequently merged into a single image. Hasselblad H4D series: the model H4D-200MS contains a 50 Mpixel sensor. The sensor sports 3 different pixel shift modes which take (a) a sequence of 6 shots taken at slight offsets, (b) a sequence of 4 shots between which the sensor is shifted by 1 pixel, (c) a sequence of 4 shots between which the sensor is shifted by 0.5 pixels. Images obtained by all three modes are subsequently merged into 200 Mpixel images. Hasselblad H5D series: both models H5D-50c MS and H5D-200c MS contain a 50 Mpixel sensor. This sensor sports 2 different pixel shift modes which take (a) a sequence of 6 shots with full and half pixel moveme
Read more →
Kernel embedding of distributions

In machine learning, the kernel embedding of distributions (also called the kernel mean or mean map) comprises a class of nonparametric methods in which a probability distribution is represented as an element of a reproducing kernel Hilbert space (RKHS). A generalization of the individual data-point feature mapping done in classical kernel methods, the embedding of distributions into infinite-dimensional feature spaces can preserve all of the statistical features of arbitrary distributions, while allowing one to compare and manipulate distributions using Hilbert space operations such as inner products, distances, projections, linear transformations, and spectral analysis. This learning framework is very general and can be applied to distributions over any space Ω {\displaystyle \Omega } on which a sensible kernel function (measuring similarity between elements of Ω {\displaystyle \Omega } ) may be defined. For example, various kernels have been proposed for learning from data which are: vectors in R d {\displaystyle \mathbb {R} ^{d}} , discrete classes/categories, strings, graphs/networks, images, time series, manifolds, dynamical systems, and other structured objects. The theory behind kernel embeddings of distributions has been primarily developed by Alex Smola, Le Song, Arthur Gretton, and Bernhard Schölkopf. A review of recent works on kernel embedding of distributions can be found in. The analysis of distributions is fundamental in machine learning and statistics, and many algorithms in these fields rely on information theoretic approaches such as entropy, mutual information, or Kullback–Leibler divergence. However, to estimate these quantities, one must first either perform density estimation, or employ sophisticated space-partitioning/bias-correction strategies which are typically infeasible for high-dimensional data. Commonly, methods for modeling complex distributions rely on parametric assumptions that may be unfounded or computationally challenging (e.g. Gaussian mixture models), while nonparametric methods like kernel density estimation (Note: the smoothing kernels in this context have a different interpretation than the kernels discussed here) or characteristic function representation (via the Fourier transform of the distribution) break down in high-dimensional settings. Methods based on the kernel embedding of distributions sidestep these problems and also possess the following advantages: Data may be modeled without restrictive assumptions about the form of the distributions and relationships between variables Intermediate density estimation is not needed Practitioners may specify the properties of a distribution most relevant for their problem (incorporating prior knowledge via choice of the kernel) If a characteristic kernel is used, then the embedding can uniquely preserve all information about a distribution, while thanks to the kernel trick, computations on the potentially infinite-dimensional RKHS can be implemented in practice as simple Gram matrix operations Dimensionality-independent rates of convergence for the empirical kernel mean (estimated using samples from the distribution) to the kernel embedding of the true underlying distribution can be proven. Learning algorithms based on this framework exhibit good generalization ability and finite sample convergence, while often being simpler and more effective than information theoretic methods Thus, learning via the kernel embedding of distributions offers a principled drop-in replacement for information theoretic approaches and is a framework which not only subsumes many popular methods in machine learning and statistics as special cases, but also can lead to entirely new learning algorithms. == Definitions == Let X {\displaystyle X} denote a random variable with domain Ω {\displaystyle \Omega } and distribution P {\displaystyle P} . Given a symmetric, positive-definite kernel k : Ω × Ω → R {\displaystyle k:\Omega \times \Omega \rightarrow \mathbb {R} } the Moore–Aronszajn theorem asserts the existence of a unique RKHS H {\displaystyle {\mathcal {H}}} on Ω {\displaystyle \Omega } (a Hilbert space of functions f : Ω → R {\displaystyle f:\Omega \to \mathbb {R} } equipped with an inner product ⟨ ⋅ , ⋅ ⟩ H {\displaystyle \langle \cdot ,\cdot \rangle _{\mathcal {H}}} and a norm ‖ ⋅ ‖ H {\displaystyle \|\cdot \|_{\mathcal {H}}} ) for which k {\displaystyle k} is a reproducing kernel, i.e., in which the element k ( x , ⋅ ) {\displaystyle k(x,\cdot )} satisfies the reproducing property ⟨ f , k ( x , ⋅ ) ⟩ H = f ( x ) ∀ f ∈ H , ∀ x ∈ Ω . {\displaystyle \langle f,k(x,\cdot )\rangle _{\mathcal {H}}=f(x)\qquad \forall f\in {\mathcal {H}},\quad \forall x\in \Omega .} One may alternatively consider x ↦ k ( x , ⋅ ) {\displaystyle x\mapsto k(x,\cdot )} as an implicit feature mapping φ : Ω → H {\displaystyle \varphi :\Omega \rightarrow {\mathcal {H}}} (which is therefore also called the feature space), so that k ( x , x ′ ) = ⟨ φ ( x ) , φ ( x ′ ) ⟩ H {\displaystyle k(x,x')=\langle \varphi (x),\varphi (x')\rangle _{\mathcal {H}}} can be viewed as a measure of similarity between points x , x ′ ∈ Ω . {\displaystyle x,x'\in \Omega .} While the similarity measure is linear in the feature space, it may be highly nonlinear in the original space depending on the choice of kernel. === Kernel embedding === The kernel embedding of the distribution P {\displaystyle P} in H {\displaystyle {\mathcal {H}}} (also called the kernel mean or mean map) is given by: μ X := E [ k ( X , ⋅ ) ] = E [ φ ( X ) ] = ∫ Ω φ ( x ) d P ( x ) {\displaystyle \mu _{X}:=\mathbb {E} [k(X,\cdot )]=\mathbb {E} [\varphi (X)]=\int _{\Omega }\varphi (x)\ \mathrm {d} P(x)} If P {\displaystyle P} allows a square integrable density p {\displaystyle p} , then μ X = E k p {\displaystyle \mu _{X}={\mathcal {E}}_{k}p} , where E k {\displaystyle {\mathcal {E}}_{k}} is the Hilbert–Schmidt integral operator. A kernel is characteristic if the mean embedding μ : { family of distributions over Ω } → H {\displaystyle \mu :\{{\text{family of distributions over }}\Omega \}\to {\mathcal {H}}} is injective. Each distribution can thus be uniquely represented in the RKHS and all statistical features of distributions are preserved by the kernel embedding if a characteristic kernel is used. === Empirical kernel embedding === Given n {\displaystyle n} training examples { x 1 , … , x n } {\displaystyle \{x_{1},\ldots ,x_{n}\}} drawn independently and identically distributed (i.i.d.) from P , {\displaystyle P,} the kernel embedding of P {\displaystyle P} can be empirically estimated as μ ^ X = 1 n ∑ i = 1 n φ ( x i ) {\displaystyle {\widehat {\mu }}_{X}={\frac {1}{n}}\sum _{i=1}^{n}\varphi (x_{i})} === Joint distribution embedding === If Y {\displaystyle Y} denotes another random variable (for simplicity, assume the co-domain of Y {\displaystyle Y} is also Ω {\displaystyle \Omega } with the same kernel k {\displaystyle k} which satisfies ⟨ φ ( x ) ⊗ φ ( y ) , φ ( x ′ ) ⊗ φ ( y ′ ) ⟩ = k ( x , x ′ ) k ( y , y ′ ) {\displaystyle \langle \varphi (x)\otimes \varphi (y),\varphi (x')\otimes \varphi (y')\rangle =k(x,x')k(y,y')} ), then the joint distribution P ( x , y ) ) {\displaystyle P(x,y))} can be mapped into a tensor product feature space H ⊗ H {\displaystyle {\mathcal {H}}\otimes {\mathcal {H}}} via C X Y = E [ φ ( X ) ⊗ φ ( Y ) ] = ∫ Ω × Ω φ ( x ) ⊗ φ ( y ) d P ( x , y ) {\displaystyle {\mathcal {C}}_{XY}=\mathbb {E} [\varphi (X)\otimes \varphi (Y)]=\int _{\Omega \times \Omega }\varphi (x)\otimes \varphi (y)\ \mathrm {d} P(x,y)} By the equivalence between a tensor and a linear map, this joint embedding may be interpreted as an uncentered cross-covariance operator C X Y : H → H {\displaystyle {\mathcal {C}}_{XY}:{\mathcal {H}}\to {\mathcal {H}}} from which the cross-covariance of functions f , g ∈ H {\displaystyle f,g\in {\mathcal {H}}} can be computed as Cov ⁡ ( f ( X ) , g ( Y ) ) := E [ f ( X ) g ( Y ) ] − E [ f ( X ) ] E [ g ( Y ) ] = ⟨ f , C X Y g ⟩ H = ⟨ f ⊗ g , C X Y ⟩ H ⊗ H {\displaystyle \operatorname {Cov} (f(X),g(Y)):=\mathbb {E} [f(X)g(Y)]-\mathbb {E} [f(X)]\mathbb {E} [g(Y)]=\langle f,{\mathcal {C}}_{XY}g\rangle _{\mathcal {H}}=\langle f\otimes g,{\mathcal {C}}_{XY}\rangle _{{\mathcal {H}}\otimes {\mathcal {H}}}} Given n {\displaystyle n} pairs of training examples { ( x 1 , y 1 ) , … , ( x n , y n ) } {\displaystyle \{(x_{1},y_{1}),\dots ,(x_{n},y_{n})\}} drawn i.i.d. from P {\displaystyle P} , we can also empirically estimate the joint distribution kernel embedding via C ^ X Y = 1 n ∑ i = 1 n φ ( x i ) ⊗ φ ( y i ) {\displaystyle {\widehat {\mathcal {C}}}_{XY}={\frac {1}{n}}\sum _{i=1}^{n}\varphi (x_{i})\otimes \varphi (y_{i})} === Conditional distribution embedding === Given a conditional distribution P ( y ∣ x ) , {\displaystyle P(y\mid x),} one can define the corresponding RKHS embedding as μ Y ∣ x = E [ φ ( Y ) ∣ X ] = ∫ Ω φ ( y ) d P ( y ∣ x ) {\displaystyle \mu _{Y\mid x}=\mathbb {E} [\varphi (Y)\mid X]=\int _{\Omega
Read more →
Logogen model

The logogen model of 1969 is a model of speech recognition that uses units called "logogens" to explain how humans comprehend spoken or written words. Logogens are a vast number of specialized recognition units, each able to recognize one specific word. This model provides for the effects of context on word recognition. == Overview == The word logogen can be traced back to the Greek-language word logos, which means "word", and genus, which means "birth". British scientist John Morton's logogen model was designed to explain word recognition using a new type of unit known as a logogen. A critical element of this theory is the involvement of lexicons, or specialized aspects of memory that include semantic and phonemic information about each item that is contained in memory. A given lexicon consists of many smaller, abstract items known as logogens. Logogens contain a variety of properties about given word such as their appearance, sound, and meaning. Logogens do not store words within themselves, but rather they store information that is specifically necessary for retrieval of whatever word is being searched for. A given logogen will become activated by psychological stimuli or contextual information (words) that is consistent with the properties of that specific logogen and when the logogen's activation level rises to or above its threshold level, the pronunciation of the given word is sent to the output system. Certain stimuli can affect the activation levels of more than one word at a time, usually involving words that are similar to one another. When this occurs, whichever of the words' activation levels reaches the threshold level, it is that word that is then sent to the output system with the subject remaining unaware of any partially excited logogens. This assumption was made by Marslen-Wilson and Welch (1978), who added to the model some assumptions of their own in order to account for their experimental results. They also assumed that the analysis of phonetic input can only become available to other parts of the system by process of how the input affects the logogen system. Finally, Marslen-Wilson and Welch assume that the first syllable of a given word will increase the activation level of a given logogen more than those of the latter syllables, which supported the data found at the time. == Analysis == The logogen model can be used to help linguists explain particular occurrences in the human language. The most-helpful application of the model is to show how one accesses words and their meanings in the lexicon. The word-frequency effect is best explained by the logogen model in that words (or logogens) that have a higher frequency (or are more common) have a lower threshold. This means that they require less perceptual power in the brain to be recognized and decoded from the lexicon and are recognized faster than those words that are less common. Also, with high-frequency words, the recovery from lowering the item's threshold is less fulfilled compared to low-frequency words so less sensory information is needed for that particular item's recognition. There are ways to lower thresholds, such as repetition and semantic priming. Also, each time a word is encountered through these methods, the threshold for that word is temporarily lowered partially because of its recovering ability. This model also conveys that specific concrete words are recalled better because they use images and logogens, whereas abstract words are not as easily recalled well because they only use logogens, hence showing the difference in thresholds between these two types of words. At the time of its conception, Morton's logogen model was one of the most influential models in springing up other parallel word access models and served as the essential basis for these subsequent models. Morton's model also strongly influenced other contemporary theories on lexical access. However, despite the advantages that the logogen theory presents, it also displays some negative facets. First and foremost, the logogen model does not explain all occurrences in language, such as the introduction of new words or non-words into a person's lexicon. Also, because of the distinctive model application, it may vary in its effectiveness in different languages. == Criticisms == While this model does a reasonable job of understanding the underlying semantics of many aspects in psycholinguistics, there are some flaws that have been pointed out in the logogen model. It has been argued that the prior stimulus patterns that have been seen in the logogen theory are not centrally localized in the logogen itself but are actually distributed throughout the different pathways over which the stimulus is being processed. What this directs at is that the notion and proliferation of logogens was due to modality. In essence, the logogen is unnecessary in the idea of attaining the title of being a recognition unit because of the variety of pathways that it is open to, not just logogens. Another criticism has been that this model essentially ignores larger and more critical structures in language and phonetics such as the different syntactic rules or grammatical construction that innately exists in language. Since this model overtly limits itself to the scope of lexical access then this model is seen as biased and misunderstood. To many psychologists, the logogen model does not meet the functional or representational adequacy that a theory should include to sufficiently comprehend language. Also, another criticism is that the logogen theory was supposed to predict that stimulus degradation should affect priming and word frequency in humans. However, many psychologists have conducted studies and researched the model to show that only priming and not word frequency is interacted with stimulus degradation. Priming is supposed to deteriorate a stimulus because it postulates that the semantic characteristics of previously known words are fed back into the detector of a person which in turn raises the threshold of related items. In word frequency, stimulus degradation is supposed to occur because it postulates that familiar words have lower thresholds than their low-frequency counterparts. However, in studies, priming is the only structure that does show observable and notable stimulus decadence. Even though the logogen theory has many unfilled holes, Morton was a revolutionary of his field whose speculation and research has opened up a remarkable era of psycholinguistics. == Other models to consider == cohort model – This model was proposed by Marslen-Wilson and was designed specifically to account for auditory word recognition. It works by breaking the word down and states that when a word is heard all words that begin with the first sound of the target word are activated. This set of words is considered the cohort. Once the first cohort has been activated, the other information, or sounds in the word narrow down the choices. The person recognizes the word when you are left with a single choice; this is considered the "recognition point". checking model – This model was developed by Norris in 1986. In this particular model, he took the approach that any word that partially matches the input is analyzed and checked to see if it fits with the context of the situation. interactive-activation model – This model is considered a connectionist model. Proposed by McClelland and Rumelhart in the 1981 to 1982 period, it is based around nodes, which are visual features, and positions of letters within a given word. They also act as word detectors which have inhibitory and excitatory connections between them. This model starts with first letter and suggests that all the words with that first letter are activated at first and then going through the word one can determine what the word is they are looking at. The main principle is that mental phenomena can be described by interconnected networks of simple units. verification model – The model was developed by Curtis Becker in 1970. The main idea is that a small number of candidates that are activated in parallel are subject to a serial-verification process. This model starts the word-recognition process with a basic representation of the stimulus. Then, sensory trace, consisting of line features is used to activate word detectors. When an acceptable number of detectors are activated these are used to generate a search set. These items are drawn from the lexicon on the basis of similarity to the sensory trace, which help with the identity of the stimulus. Then, in a serial process the candidates are compared to the representation of the sensory-trace input. == Related concepts == word frequency – This is the belief that the speed and accuracy with which a word is recognized is related to how frequently the word occurs in our language. Each logogen has a threshold (for identification) and words with higher frequencies have lower thresholds. Words with higher freq
Read more →
Image warping

Image warping is the process of digitally manipulating an image such that any shapes portrayed in the image have been significantly distorted. Warping may be used for correcting image distortion as well as for creative purposes (e.g., morphing). The same techniques are equally applicable to video. While an image can be transformed in various ways, pure warping means that points are mapped to points without changing the colors. This can be based mathematically on any function from (part of) the plane to the plane. If the function is injective the original can be reconstructed. If the function is a bijection any image can be inversely transformed. Some methods are: Images may be distorted through simulation of optical aberrations. Images may be viewed as if they had been projected onto a curved or mirrored surface. (This is often seen in ray traced images.) Images can be partitioned into image polygons and each polygon distorted. Images can be distorted using morphing. The most obvious approach to transforming a digital image is the forward mapping. This applies the transform directly to the source image, typically generating unevenly-spaced points that will then be interpolated to generate the required regularly-spaced pixels. However, for injective transforms reverse mapping is also available. This applies the inverse transform to the target pixels to find the unevenly-spaced locations in the source image that contribute to them. Estimating them from source image pixels will require interpolation of the source image. To work out what kind of warping has taken place between consecutive images, one can use optical flow estimation techniques. == Image warping toolbox == ImWIP is an open-source, image warping tool for modeling deformation and motion in digital images, which contains differentiable image warping operators, together with their exact adjoints and derivatives.
Read more →
Speech recognition

Speech recognition (automatic speech recognition (ASR), computer speech recognition, or speech-to-text (STT)) is a sub-field of computational linguistics concerned with methods and technologies that translate spoken language into text or other interpretable forms. Speech recognition applications include voice user interfaces, where the user speaks to a device, which "listens" and processes the audio. Common voice applications include interpreting commands for calling, call routing, home automation, and aircraft control. These applications are called direct voice input. Productivity applications include searching audio recordings, creating transcripts, and dictation. Speech recognition can be used to analyse speaker characteristics, such as identifying native language using pronunciation assessment. Voice recognition (speaker identification) refers to identifying the speaker, rather than speech contents. Recognizing the speaker can simplify the task of translating speech in systems trained on a specific person's voice. It can also be used to authenticate the speaker as part of a security process. == History == Applications for speech recognition developed over many decades, with progress accelerated due to advances in deep learning and the use of big data. These advances are reflected in an increase in academic papers, and greater system adoption. Key areas of growth include vocabulary size, more accurate recognition for unfamiliar speakers (speaker independence), and faster processing speed. === Pre-1970 === 1952 – Bell Labs researchers, Stephen Balashek, R. Biddulph, and K. H. Davis, built Audrey for single-speaker digit recognition. Their system located the formants in the power spectrum of each utterance. 1960 – Gunnar Fant developed and published the source–filter model of speech production. 1962 – IBM's 16-word "Shoebox" machine's speech recognition debuted at the 1962 World's Fair. 1966 – Linear predictive coding, a speech coding method, was proposed by Fumitada Itakura of Nagoya University and Shuzo Saito of Nippon Telegraph and Telephone. 1969 – Funding at Bell Labs came to a halt for several years after the company's head engineer, John R. Pierce, wrote an open letter criticizing speech recognition research. This defunding lasted until Pierce retired and James L. Flanagan took over. Raj Reddy was the first person to work on continuous speech recognition, as a graduate student at Stanford University in the late 1960s. Previous systems required users to pause after each word. Reddy's system issued spoken commands for playing chess. Around this time, Soviet researchers invented the dynamic time warping (DTW) algorithm and used it to create a recognizer capable of operating on a 200-word vocabulary. DTW processed speech by dividing it into short frames (e.g. 10 ms segments) and treating each frame as a unit. Speaker independence, however, remained unsolved. === 1970–1990 === 1971 – DARPA funded a five-year speech recognition research project, Speech Understanding Research, seeking a minimum vocabulary size of 1,000 words. The project considered speech understanding a key to achieving progress in speech recognition, which was later disproved. BBN, IBM, Carnegie Mellon (CMU), and Stanford Research Institute participated. 1972 – The IEEE Acoustics, Speech, and Signal Processing group held a conference in Newton, Massachusetts. 1976 – The first ICASSP was held in Philadelphia, which became a major venue for publishing on speech recognition. During the late 1960s, Leonard Baum developed the mathematics of Markov chains at the Institute for Defense Analysis. A decade later, at CMU, Raj Reddy's students James Baker and Janet M. Baker began using the hidden Markov model (HMM) for speech recognition. James Baker had learned about HMMs while at the Institute for Defense Analysis. HMMs enabled researchers to combine sources of knowledge, such as acoustics, language, and syntax, in a unified probabilistic model. By the mid-1980s, Fred Jelinek's team at IBM created a voice-activated typewriter called Tangora, which could handle a 20,000-word vocabulary. Jelinek's statistical approach placed less emphasis on emulating human brain processes in favor of statistical modelling. (Jelinek's group independently discovered the application of HMMs to speech.) This was controversial among linguists since HMMs are too simplistic to account for many features of human languages. However, the HMM proved to be a highly useful way for modelling speech and replaced dynamic time warping as the dominant speech recognition algorithm in the 1980s. 1982 – Dragon Systems, founded by James and Janet M. Baker, was one of IBM's few competitors. === Practical speech recognition === The 1980s also saw the introduction of the n-gram language model. 1987 – The back-off model enabled language models to use multiple-length n-grams, and CSELT used HMM to recognize languages (in software and hardware, e.g. RIPAC). At the end of the DARPA program in 1976, the best computer available to researchers was the PDP-10 with 4 MB of RAM. It could take up to 100 minutes to decode 30 seconds of speech. Practical products included: 1984 – the Apricot Portable was released with up to 4096 words support, of which only 64 could be held in RAM at a time. 1987 – a recognizer from Kurzweil Applied Intelligence 1990 – Dragon Dictate, a consumer product released in 1990. AT&T deployed the Voice Recognition Call Processing service in 1992 to route telephone calls without a human operator. The technology was developed by Lawrence Rabiner and others at Bell Labs. By the early 1990s, the vocabulary of the typical commercial speech recognition system had exceeded the average human vocabulary. Reddy's former student, Xuedong Huang, developed the Sphinx-II system at CMU. Sphinx-II was the first to do speaker-independent, large vocabulary, continuous speech recognition, and it won DARPA's 1992 evaluation. Handling continuous speech with a large vocabulary was a major milestone. Huang later founded the speech recognition group at Microsoft in 1993. Reddy's student Kai-Fu Lee joined Apple, where, in 1992, he helped develop the Casper speech interface prototype. Lernout & Hauspie, a Belgium-based speech recognition company, acquired other companies, including Kurzweil Applied Intelligence in 1997 and Dragon Systems in 2000. L&H was used in Windows XP. L&H was an industry leader until an accounting scandal destroyed it in 2001. L&H speech technology was bought by ScanSoft, which became Nuance in 2005. Apple licensed Nuance software for its digital assistant Siri. ==== 2000s ==== In the 2000s, DARPA sponsored two speech recognition programs: Effective Affordable Reusable Speech-to-Text (EARS) in 2002, followed by Global Autonomous Language Exploitation (GALE) in 2005. Four teams participated in EARS: IBM; a team led by BBN with LIMSI and the University of Pittsburgh; Cambridge University; and a team composed of ICSI, SRI, and the University of Washington. EARS funded the collection of the Switchboard telephone speech corpus, which contained 260 hours of recorded conversations from over 500 speakers. The GALE program focused on Arabic and Mandarin broadcast news. Google's first effort at speech recognition came in 2007 after recruiting Nuance researchers. Its first product, GOOG-411, was a telephone-based directory service. Since at least 2006, the U.S. National Security Agency has employed keyword spotting, allowing analysts to index large volumes of recorded conversations and identify speech containing "interesting" keywords. Other government research programs focused on intelligence applications, such as DARPA's EARS program and IARPA's Babel program. In the early 2000s, speech recognition was dominated by hidden Markov models combined with feed-forward artificial neural networks (ANN). Later, speech recognition was taken over by long short-term memory (LSTM), a recurrent neural network (RNN) published by Sepp Hochreiter & Jürgen Schmidhuber in 1997. LSTM RNNs avoid the vanishing gradient problem and can learn "Very Deep Learning" tasks that require memories of events that happened thousands of discrete time steps earlier, which is important for speech. Around 2007, LSTMs trained with Connectionist Temporal Classification (CTC) began to outperform. In 2015, Google reported a 49 percent error‑rate reduction in its speech recognition via CTC‑trained LSTM. Transformers, a type of neural network based solely on attention, were adopted in computer vision and language modelling, and then to speech recognition. Deep feed-forward (non-recurrent) networks for acoustic modelling were introduced in 2009 by Geoffrey Hinton and his students at the University of Toronto, and by Li Deng and colleagues at Microsoft Research. In contrast to the prioer incremental improvements, deep learning decreased error rates by 30%. Both shallow and deep forms (e.g., recurrent nets) of ANNs had been explored since the 1980s. Howev
Read more →
Engineering Historical Memory

Engineering Historical Memory (EHM) is an online database in the digital humanities, serving as an open-access research tool for primary historical materials focused on 11th to 15th century Afro-Eurasia. It adopts computational methods to make historical documents machine-understandable. EHM parses traditional artifacts such as historical maps, travel accounts, chronicles and codices into computer-readable formats, and links them to secondary multi-media references, a process referred to as the "automatic narrative generation". This approach generates cultural narratives and facilitates interaction with the historical artifacts, making them accessible to audiences from various backgrounds. == History == EHM was first theorised in 2007 by researcher Andrea Nanetti when he was a visiting scholar at Princeton University, and the preliminary test results were published between 2008 and 2011. In 2013, the EHM research team was set up in Singapore following Nanetti's professorship at Nanyang Technological University (NTU). Two years later, after receiving several Microsoft research grants, EHM went live on Microsoft Azure. In 2018, the College of Humanities, Arts and Social Sciences (CoHASS) at NTU Singapore formed the Digital Humanities Research Cluster, as part of which, EHM has been an ongoing interdisciplinary research project led by Nanetti. Partnering with international educational and cultural institutions such as Ca' Foscari University of Venice, University of Florence, Taylor & Francis Group, Delft University of Technology (TUDelft), and SenticNet, EHM has been supported by over 130 scholars and engineers. == Applications == Primary historical materials on EHM are curated into several categories, including maps, travel accounts, chronicles, codices, sites, archival documents, and paintings, such as the Morosini Codex (listed under Chronicles) and Pope Gregory X's Privilege for the Holy Monastery of St Catherine of Sinai (listed under Archival Documents). EHM has been adopted by cultural organisations as an exhibition and research tool in the digital humanities field. An example is the publication of a digital interactive edition of Fra Mauro's Map of the World on EHM, a collaboration project between NTU Singapore and the Biblioteca Nazionale Marciana of Venice. The digitisation process of the map on EHM involved transcribing and geo-referencing the textual content in the 15th-century map, followed by creating semantic annotations to connect the map's content with related secondary data sources. The e-map was subsequently adopted and launched online by Museo Galileo in March 2022 and incorporated into the virtual exhibition "Venezia and Suzhou: Water Cities along the Silk Roads" (online, September-December 2022). In 2024, the Fra Mauro's Map of the World application on EHM was awarded the Digital Humanities and Multimedia Studies Prize (DHMS) by the Medieval Academy of America. Image-Based Video Search Engine is another experimental project under the EHM scope led by the research teams at Delft University of Technology (TUDelft) and NTU Singapore. This ongoing project aims to improve the efficiency of retrieving targeted objects from audio-visuals. == Awards == In 2021, EHM won the GLAMi Awards (MuseWeb Conference - Galleries, Libraries, Archives, and Museums Innovation awards) in the "Resources for Scholars and Researchers" category. In the same year, EHM was a Falling Walls finalist for Science Breakthrough of the Year in the category Social Sciences and Humanities after nominated by the School of Advanced Study at the University of London. In April 2022, the Italian National Commission for UNESCO has selected and sent the EHM project to the organisers of the "Jikji Memory of the World" Award for final evaluation. In January 2024, the Medieval Academy of America announced its 2024 Digital Humanities and Multimedia Studies Prize (DHMS) goes to the Fra Mauro's Map of the World application on EHM.
Read more →
Video editing software

Video editing software or a video editor is software used for performing the post-production video editing of digital video sequences on a non-linear editing system (NLE). It has replaced traditional flatbed celluloid film editing tools and analog video tape editing machines. Video editing software serves a lot of purposes, such as filmmaking, audio commentary, and general editing of video content. In NLE software, the user manipulates sections of video, images, and audio on a sequence. These clips can be trimmed, cut, and manipulated in many different ways. When editing is finished, the user exports the sequence as a video file. == Components == === Timeline === NLE software is typically based on a timeline interface where sections moving image video recordings, known as clips, are laid out in sequence and played back. The NLE offers a range of tools for trimming, splicing, cutting, and arranging clips across the timeline. Another kind of clip is a text clip, used to add text to a video, such as title screens or movie credits. Audio clips can additionally be mixed together, such as mixing a soundtrack with multiple sound effects. Typically, the timeline is divided into multiple rows on the y-axis for different clips playing simultaneously, whereas the x-axis represents the run time of the video. Effects such as transitions can be performed on each clip, such as a crossfade effect going from one scene to another. === Exporting === Since video editors represent a project with a file format specific to the program, one needs to export the video file in order to publish it. Once a project is complete, the editor can then export to movies in a variety of formats in a context that may range from broadcast tape formats to compressed video files for web publishing (such as on an online video platform or personal website), optical media, or saved to mobile devices. To facilitate editing, source video typically has a higher resolution than the desired output. Therefore, higher resolution video needs to be downscaled during exporting, or after exporting in a process known as transsizing. === Visual effects === As digital video editing advanced, visual effects became possible, and is part of the standard toolkit, usually found in prosumer and professional grade software. A common ability is to do compositing techniques such as chroma keying or luma keying, among others, which allow different objects to look as if they are in the same scene. A different kind of visual effects is motion capture. Software such as Blender can perform motion capture to make animated objects follow an actor's movements. === Additional features === Most professional video editors are able to do color grading, which is to manipulate visual attributes of a video such as contrast to enhance output, and improve emotional impact. Some video editors such as iMovie include stock footage available for use. == Hardware requirements == As video editing puts great demands on storage and graphics performance, especially at high resolutions such as 4K, and for videos with many visual effects, powerful hardware is often required. It is not uncommon for a computer built for video editing to have a lot of drive capacity, and a powerful graphics processing unit, which optimally has hardware accelerated video encoding. Having sufficient disk space is important since videos can take up large amounts of storage, depending on the resolution and compression format used. Each minute of a Full HD (1080p) video at 30 fps takes up 60MB of space. When visual effects are used, a server farm can be employed to speed up the rendering process. == Examples == Video editing software can be divided into consumer grade, which focuses on ease-of-use, along with professional grade software, which focuses on feature availability, and advanced editing techniques. The typical use case for the former is to edit personal videos on the go, when more advanced editing is not required. === Consumer grade === Photos (Apple) Google Photos YouTube Create === Prosumer grade === ==== Proprietary software ==== iMovie CyberLink PowerDirector === Professional grade === ==== Proprietary software ==== Final Cut Pro Adobe Premiere Pro DaVinci Resolve Vegas Pro Lightworks Camtasia Media Composer ==== Free and open source software ==== Avidemux Blender Cinelerra Flowblade Kdenlive OpenShot Shotcut While most video editing software has been separate from the operating systems, some operating systems have had a video editor installed by default, such as Windows Movie Maker in Windows XP, or as a component of the default photo viewer, such as the Photos app on iOS. Some social media platforms, such as TikTok and Instagram may include a rudimentary video editor to trim clips.
Read more →
Apptek

Applications Technology (AppTek) is a U.S. company headquartered in McLean, Virginia that specializes in artificial intelligence and machine learning for human language technologies. The company provides both managed and professional services for natural language processing (NLP) technologies including automatic speech recognition (ASR), neural machine translation (MT), natural-language understanding (NLU) and neural speech synthesis. AppTek's Head of Science, Prof. Dr. -Ing Hermann Ney, was awarded the IEEE James L. Flanagan Speech and Audio Processing Award in 2019 and the ISCA Medal for Scientific Achievement in 2021 for his work in natural language processing. == History == AppTek was acquired in 1998 by Lernout & Hauspie (at the time a NASDAQ publicly traded company), AppTek organized a management buy-out and went private again in 2001. In 2014, the company sold its hybrid machine translation technology to eBay and has since rebuilt the platform to modern neural-based approaches for machine translation. In 2020, SOSi acquired non-controlling interest in AppTek and became an exclusive reseller of AppTek products for U.S. federal, state, and local government entities.
Read more →