Josh (stylized as JOSH) was a video-sharing social networking service but it has since evolved into a live call and chat application owned by VerSe Innovation – an Indian technology company based in Bangalore, India. Josh was an Indian short video app that was launched in immediately after the Indian Government banned TikTok and other Chinese apps in June 2020. The founders of the platform have promoted the app as the “Instagram for Bharat” referring to their focus on the Indian audience that speaks its own regional and state languages. Josh was among the top 10 most downloaded apps social and entertainment apps in India of 2021 and had 150 million monthly active users as per April 2022. The word 'Josh' translates to fervour or passion. The app was launched under the aegis of the Atmanirbhar Bharat campaign and to compete with the duopoly of Google and Facebook in India. Josh's parent company VerSe Innovations Pvt. Ltd. owns another startup Dailyhunt, which a content and news aggregator application. Both Dailyhunt and Josh are a part of the VerSe's focus on the "next billion" regional language users of India. Founders Virendra Gupta and Umang Bedi conceptualised Josh as a short-video platform that made content creation accessible to vernacular language users, essentially the non-English speaking audience in India. == Features == Josh is currently available in 12 Indian languages and allows users to upload, share, remix bite-sized videos of up to 120 seconds. There are various categories across the video section including viral, trending, glamour, dance, devotion, yoga and cooking among others. Similar to Instagram and TikTok, it has a video feed which is curated for individuals on the basis of their app behaviour. The app hosts many daily, weekly and monthly social media challenges. == Funding == In December 2020, within 3 months of its launch, Josh's parent app VerSe Innovation raised more than $100 million from investors including Alphabet Inc's Google and Microsoft. In February 2021, VerSe Innovation raised $100 million in Series H funding from Qatar Investment Authority, the sovereign wealth fund of the State of Qatar, and Glade Brook Capital Partners. In August 2021, VerSe raised over $450 million in its Series I financing round with a valuation of $1 billion. Investors included Canada Pension Plan Investment Board (CPPIB), Siguler Guff, Baillie Gifford, Carlyle Asia Partners Growth II affiliates, and others. The startup announced its plan to expand overseas and broaden its ecommerce play for both Dailyhunt and Josh. In April 2022, VerSe announced that it has raised $805 million in funding from investors at a valuation of nearly $5 billion. ByteDance Offloads Stake In Josh Parent VerSe, Exits At 56% Discount == Partnerships == In February 2021, Saregama and Josh signed a music licensing deal, wherein Josh expanded its musical library with 1.3 lakh songs from Saregama in 25 different languages. To improve their user experience, Josh partnered with computer vision company D-ID in August 2021. The company helped Josh introduce photo-to-video features, live portrait technology, animate their photos etc. In order to solidify their efforts in enhancing Josh, VerSe acquired Indian social networking platform GolBol in October 2021. The move came as an effort by the startup to strengthen their discovery initiatives on the platform and classify content at scale and understand the core behaviour of Indian regional audiences. Josh has also announced its plans to include live commerce as a potential revenue stream through its partnership with multiple large e-commerce players. == Notable campaigns == Say No To Dowry – In association with Josh, the Kerala Police partook in the #SayNo2Dowry online social media campaign that was started to highlight and stop the social evil in the state. Salute India – Josh entered the Guinness World Records by creating the largest online video album of people saluting (29,529). It organised an online campaign #SaluteIndia on the app during the 75th Independence Day of India during 10–15 August 2021.
Eigenface
An eigenface ( EYE-gən-) is the name given to a set of eigenvectors when used in the computer vision problem of human face recognition. The approach of using eigenfaces for recognition was developed by Sirovich and Kirby and used by Matthew Turk and Alex Pentland in face classification. The eigenvectors are derived from the covariance matrix of the probability distribution over the high-dimensional vector space of face images. The eigenfaces themselves form a basis set of all images used to construct the covariance matrix. This produces dimension reduction by allowing the smaller set of basis images to represent the original training images. Classification can be achieved by comparing how faces are represented by the basis set. == History == The eigenface approach began with a search for a low-dimensional representation of face images. Sirovich and Kirby showed that principal component analysis could be used on a collection of face images to form a set of basis features. These basis images, known as eigenpictures, could be linearly combined to reconstruct images in the original training set. If the training set consists of M images, principal component analysis could form a basis set of N images, where N < M. The reconstruction error is reduced by increasing the number of eigenpictures; however, the number needed is always chosen less than M. For example, if you need to generate a number of N eigenfaces for a training set of M face images, you can say that each face image can be made up of "proportions" of all the K "features" or eigenfaces: Face image1 = (23% of E1) + (2% of E2) + (51% of E3) + ... + (1% En). In 1991 M. Turk and A. Pentland expanded these results and presented the eigenface method of face recognition. In addition to designing a system for automated face recognition using eigenfaces, they showed a way of calculating the eigenvectors of a covariance matrix such that computers of the time could perform eigen-decomposition on a large number of face images. Face images usually occupy a high-dimensional space and conventional principal component analysis was intractable on such data sets. Turk and Pentland's paper demonstrated ways to extract the eigenvectors based on matrices sized by the number of images rather than the number of pixels. Once established, the eigenface method was expanded to include methods of preprocessing to improve accuracy. Multiple manifold approaches were also used to build sets of eigenfaces for different subjects and different features, such as the eyes. == Generation == A set of eigenfaces can be generated by performing a mathematical process called principal component analysis (PCA) on a large set of images depicting different human faces. Informally, eigenfaces can be considered a set of "standardized face ingredients", derived from statistical analysis of many pictures of faces. Any human face can be considered to be a combination of these standard faces. For example, one's face might be composed of the average face plus 10% from eigenface 1, 55% from eigenface 2, and even −3% from eigenface 3. Remarkably, it does not take many eigenfaces combined together to achieve a fair approximation of most faces. Also, because a person's face is not recorded by a digital photograph, but instead as just a list of values (one value for each eigenface in the database used), much less space is taken for each person's face. The eigenfaces that are created will appear as light and dark areas that are arranged in a specific pattern. This pattern is how different features of a face are singled out to be evaluated and scored. There will be a pattern to evaluate symmetry, whether there is any style of facial hair, where the hairline is, or an evaluation of the size of the nose or mouth. Other eigenfaces have patterns that are less simple to identify, and the image of the eigenface may look very little like a face. The technique used in creating eigenfaces and using them for recognition is also used outside of face recognition: handwriting recognition, lip reading, voice recognition, sign language/hand gestures interpretation and medical imaging analysis. Therefore, some do not use the term eigenface, but prefer to use 'eigenimage'. === Practical implementation === To create a set of eigenfaces, one must: Prepare a training set of face images. The pictures constituting the training set should have been taken under the same lighting conditions, and must be normalized to have the eyes and mouths aligned across all images. They must also be all resampled to a common pixel resolution (r × c). Each image is treated as one vector, simply by concatenating the rows of pixels in the original image, resulting in a single column with r × c elements. For this implementation, it is assumed that all images of the training set are stored in a single matrix T, where each column of the matrix is an image. Subtract the mean. The average image a has to be calculated and then subtracted from each original image in T. Calculate the eigenvectors and eigenvalues of the covariance matrix S. Each eigenvector has the same dimensionality (number of components) as the original images, and thus can itself be seen as an image. The eigenvectors of this covariance matrix are therefore called eigenfaces. They are the directions in which the images differ from the mean image. Usually this will be a computationally expensive step (if at all possible), but the practical applicability of eigenfaces stems from the possibility to compute the eigenvectors of S efficiently, without ever computing S explicitly, as detailed below. Choose the principal components. Sort the eigenvalues in descending order and arrange eigenvectors accordingly. The number of principal components k is determined arbitrarily by setting a threshold ε on the total variance. Total variance v = ( λ 1 + λ 2 + . . . + λ n ) {\displaystyle v=(\lambda _{1}+\lambda _{2}+...+\lambda _{n})} , n = number of components, and λ {\displaystyle \lambda } represents component eigenvalue. k is the smallest number that satisfies ( λ 1 + λ 2 + . . . + λ k ) v > ϵ {\displaystyle {\frac {(\lambda _{1}+\lambda _{2}+...+\lambda _{k})}{v}}>\epsilon } These eigenfaces can now be used to represent both existing and new faces: we can project a new (mean-subtracted) image on the eigenfaces and thereby record how that new face differs from the mean face. The eigenvalues associated with each eigenface represent how much the images in the training set vary from the mean image in that direction. Information is lost by projecting the image on a subset of the eigenvectors, but losses are minimized by keeping those eigenfaces with the largest eigenvalues. For instance, working with a 100 × 100 image will produce 10,000 eigenvectors. In practical applications, most faces can typically be identified using a projection on between 100 and 150 eigenfaces, so that most of the 10,000 eigenvectors can be discarded. === Matlab example code === Here is an example of calculating eigenfaces with Extended Yale Face Database B. To evade computational and storage bottleneck, the face images are sampled down by a factor 4×4=16. Note that although the covariance matrix S generates many eigenfaces, only a fraction of those are needed to represent the majority of the faces. For example, to represent 95% of the total variation of all face images, only the first 43 eigenfaces are needed. To calculate this result, implement the following code: === Computing the eigenvectors === Performing PCA directly on the covariance matrix of the images is often computationally infeasible. If small images are used, say 100 × 100 pixels, each image is a point in a 10,000-dimensional space and the covariance matrix S is a matrix of 10,000 × 10,000 = 108 elements. However the rank of the covariance matrix is limited by the number of training examples: if there are N training examples, there will be at most N − 1 eigenvectors with non-zero eigenvalues. If the number of training examples is smaller than the dimensionality of the images, the principal components can be computed more easily as follows. Let T be the matrix of preprocessed training examples, where each column contains one mean-subtracted image. The covariance matrix can then be computed as S = TTT and the eigenvector decomposition of S is given by S v i = T T T v i = λ i v i {\displaystyle \mathbf {Sv} _{i}=\mathbf {T} \mathbf {T} ^{T}\mathbf {v} _{i}=\lambda _{i}\mathbf {v} _{i}} However TTT is a large matrix, and if instead we take the eigenvalue decomposition of T T T u i = λ i u i {\displaystyle \mathbf {T} ^{T}\mathbf {T} \mathbf {u} _{i}=\lambda _{i}\mathbf {u} _{i}} then we notice that by pre-multiplying both sides of the equation with T, we obtain T T T T u i = λ i T u i {\displaystyle \mathbf {T} \mathbf {T} ^{T}\mathbf {T} \mathbf {u} _{i}=\lambda _{i}\mathbf {T} \mathbf {u} _{i}} Meaning that, if ui is an eigenvector of TTT, then vi = Tui is an eigenvector of S. If we have
Outline of electronics
The following outline is provided as an overview of and topical guide to electronics: Electronics – branch of physics, engineering and technology dealing with electrical circuits that involve active semiconductor components and associated passive interconnection technologies. == Branches == === Classical electronics === Analog electronics Digital electronics Electronic instrumentation Electronic engineering Microelectronics Optoelectronics Power electronics Printed electronics Semiconductor technology Schematic capture Thermal management Automation Electronics === Advanced topics === Atomtronics Bioelectronics Failure modes of electronics Flexible electronics Low-power electronics Microelectromechanical systems (MEMS) Molecular electronics Nanoelectronics Organic electronics Photonics Piezotronics Quantum electronics Spintronics === History of electronics === History of electronic engineering History of radar History of radio History of television == General concepts == === Data converters === Analog-to-digital converters (ADC) Aliasing Successive approximation ADC Dual-slope ADC Quantization Sensor resolution Sampling Delta-sigma ADC Digital-to-analog converters (DAC) Digital potentiometer Binary weighted resistor converter Charge distribution DAC Pulse width modulator Reconstruction filter The R2R ladder === Digital electronics === Binary decision diagrams Boolean algebra Combinational logic Counters (digital) De Morgan's laws Digital circuit Formal verification Karnaugh maps Logic families Logic gate Logic minimization Logic simulation Logic synthesis Registers Sequential logic State machines Truth tables Transparent latch === Electrical element/discretes === Passive elements: Capacitor Inductor Memristor Resistor Transformer Active elements: Diode Zener diode Light-emitting diode PIN diode Schottky diode Avalanche diode Laser diode Microcontroller Operational amplifier Thyristor DIAC TRIAC IGBT Transistor Bipolar transistor (BJT) Field effect transistor (FET) Darlington transistor Other components Aural devices Battery (electricity) Crystal oscillator Electromechanical devices Sensors Surface acoustic wave (SAW) === Electronics analysis === Electronic packaging Electronic circuit simulation Electronic design automation Electronic noise Mathematical methods in electronics Thermal management of electronic devices and systems === Electronic circuits === Amplifiers Differential amplifiers Feedback amplifiers Power amplifiers Comparators Converters Filters Active filters Passive filters Digital filters Oscillators Phase-locked loops Timers === Electronic equipment === Air conditioner Breathalyzer Central heating Clothes dryer Computer/Notebook Dishwasher Freezer Home robot Home entertainment system Information technologies Cooker Microwave oven Refrigerator Robotic vacuum cleaner Tablet Telephone Water heater Washing machine === Television === Analog television History of television Television show Television broadcaster Timeline of the introduction of television in countries Mechanical television Color television Digital television Digital television transition Smart television Streaming television Internet Protocol television 3D television Terrestrial television ==== Television broadcasting ==== === Electronic instrumentation === Ammeter Capacitance meter Distortionmeter Electric energy meter LCR meter Microwave power meter Multimeter Network analyzer Ohmmeter Oscilloscope Psophometer Q meter Signal analyzer Signal generator Spectrum analyzer Transistor tester Tube tester Wattmeter Vectorscope Video signal generator Voltmeter VU meter === Memory technology === Flash memory Hard drive systems Optical storage Probe Storage Programmable read-only memory Read-only memory Solid-state drive (SSD) Volatile memory === Microcontrollers === Features Analog-to-digital converter Central processing unit (CPU) Clock generator (Quartz timing crystal, resonator or RC circuit) Debugging support Digital-to-analog converters Discrete input and output bits In-circuit programming Non-volatile memory (ROM, EPROM, EEPROM or Flash) Peripherals (Timers, event counters, PWM generators, and watchdog) Serial interface (Input/output such as serial ports (UARTs)) Serial communications (I²C, Serial Peripheral Interface and Controller Area Network) Volatile memory (RAM) 8-bit microcontroller families: AVR - PIC - COP8 - MCS-48 - MCS-51 - Z8 - eZ80 - HC08 - HC11 - H8 - PSoC Some notable suppliers: ARM Atmel Cypress Semiconductor Freescale Intel MIPS Microchip Technology NXP Semiconductors Parallax Propeller PowerPC Rabbit 2000 Renesas RX, V850 Silicon Laboratories STMicroelectronics Texas Instruments Toshiba TLCS === Optoelectronics === Optical fiber Optical properties Optical receivers Optical system design Optical transmitters === Physical laws === Ampère's law Coulomb's law Faraday's law of induction/Faraday-Lenz law Gauss's law Kirchhoff's circuit laws Current law Voltage law Maxwell's equations Gauss's law Faraday's law of induction Ampère's law Ohm's law === Power electronics === Power Devices Gate turn-off thyristor MOS-controlled thyristor (MCT) Power BJT/MOSFET Static induction devices Electric power conversion DC to DC DC to DC converter Voltage stabiliser Linear regulator AC to DC Rectifier Mains power supply unit (PSU) Switched-mode power supply DC to AC Inverter AC to AC Cycloconverter Transformer Variable frequency transformer Voltage converter Voltage regulator Power applications Automotive applications Capacitor charging applications Electronic ballasts Energy harvesting technologies Flexible AC transmission systems (FACTS) High frequency inverters HVDC transmission Motor controller Photovoltaic system Conversion Power factor correction circuits Power supply Renewable energy sources Switching power converters Uninterruptible power supply Wind power === Programmable devices === Application-specific integrated circuit (ASIC) Complex programmable logic device (CPLD) Erasable programmable logic device (EPLD) Simple programmable logic device (SPLD) Macrocell array Programmable array logic (PAL) Programmable logic array (PLA) Programmable logic device (PLD) Field-programmable gate array (FPGA) VHSIC Hardware Description Language (VHDL) Verilog Hardware Description Language Some notable suppliers: Altera - Atmel - Cypress Semiconductor - Lattice Semiconductor - Xilinx === Semiconductors theory === Properties Bipolar junction transistors Capacitance voltage profiling Charge carrier Charge-transfer complex Deep-level transient spectroscopy Depletion region Density of states Diode modelling Direct band gap Electronic band structure Energy level Exciton Field-effect transistors Metal–semiconductor junction MOSFETs N-type semiconductor Organic semiconductors P–n junction P-type semiconductor Photoelectric effect Quantum tunneling Semiconductor chip Semiconductor detector Solar cell Transistor model Thin film Tight-binding model Device Fabrication Semiconductor device fabrication Semiconductor industry Semiconductor consolidation == Applications == Audio electronics Automotive electronics Avionics Control Systems Consumer electronics Data acquisition E-health Electronic book Electronics industry Electronic warfare Embedded systems Home automation Integrated circuits Marine electronics Microwave technology Military electronics Multimedia Nuclear electronics Open hardware Radar and Radionavigation Radio electronics Terahertz technology Video hardware Wired and Wireless Communications
Digital intermediate
Digital intermediate (DI) is a motion picture finishing process which classically involves digitizing a motion picture and manipulating the color and other image characteristics. == Definition and overview == A digital intermediate often replaces or augments the photochemical timing process and is usually the final creative adjustment to a movie before distribution in theaters. It is distinguished from the telecine process in which film is scanned and color is manipulated early in the process to facilitate editing. However the lines between telecine and DI are continually blurred and are often executed on the same hardware by colorists of the same background. These two steps are typically part of the overall color management process in a motion picture at different points in time. A digital intermediate is also customarily done at higher resolution and with greater color fidelity than telecine transfers. Although originally used to describe a process that started with film scanning and ended with film recording, digital intermediate is also used to describe color correction and color grading and even final mastering when a digital camera is used as the image source and/or when the final movie is not output to film. This is due to recent advances in digital cinematography and digital projection technologies that strive to match film origination and film projection. In traditional photochemical film finishing, an intermediate is produced by exposing film to the original camera negative. The intermediate is then used to mass-produce the films that get distributed to theaters. Color grading is done by varying the amount of red, green, and blue light used to expose the intermediate. The digital intermediate process uses digital tools to color grade, which allows for much finer control of individual colors and areas of the image, and allows for the adjustment of image structure (grain, sharpness, etc.). The intermediate for film reproduction can then be produced by means of a film recorder. The physical intermediate film that is a result of the recording process is sometimes also called a digital intermediate, and is usually recorded to internegative (IN) stock, which is inherently finer-grain than original camera negative (OCN). One of the key technical achievements that made the transition to DI possible was the use of 3D look-up tables, which could be used to mimic how the digital image would look once it was printed onto release print stock. This removed a large amount of guesswork from the film-making process, and allowed greater freedom in the colour grading process while reducing risk. The digital master is often used as a source for a DCI-compliant distribution of the motion picture for digital projection. For archival purposes, the digital master created during the digital intermediate process can be recorded to very stable high dynamic range yellow-cyan-magenta (YCM) separations on black-and-white film with an expected 100-year or longer life. While still subject to the natural degradation of any analog chemical master, this archival format, long used in the industry prior to the invention of DI, was considered valuable for providing an archival medium that is independent of changes in digital data recording technologies and file formats that might otherwise render digitally archived material unreadable in the long term. A "film intermediate" is an analog variation of a digital intermediate, where a project shot on digital video is printed onto film stock and transferred back to digital video to emulate film. The term was coined after it was used on the Oscar-winning 2012 short film "Curfew". The process was also used on the films Dune (2021) and The Batman (2022). == History == Telecine tools to electronically capture film images are nearly as old as broadcast television, but the resulting images were widely considered unsuitable for exposing back onto film for theatrical distribution. Film scanners and recorders with quality sufficient to produce images that could be inter-cut with regular film began appearing in the 1970s, with significant improvements in the late 1980s and early 1990s. During this time, digitally processing an entire feature-length film was impractical because the scanners and recorders were extremely slow and the image files were too large compared to computing power available. Instead, individual shots or short sequences were processed for visual effects. In 1992, Visual Effects Supervisor/Producer Chris F. Woods broke through several "techno-barriers" in creating a digital studio to produce the visual effects for the 1993 release Super Mario Bros. It was the first feature film project to digitally scan a large number of VFX plates (over 700) at 2K resolution. It was also the first film scanned and recorded at Kodak's just launched Cinesite facility in Hollywood. This project based studio was the first feature film to use Discreet Logic's (now Autodesk) Flame and Inferno systems, which enjoyed early dominance as high resolution / high performance digital compositing systems. Digital film compositing for visual effects was immediately embraced, while optical printer use for VFX declined just as quickly. Chris Watts further revolutionized the process on the 1998 feature film Pleasantville, becoming the first visual effects supervisor for New Line Cinema to scan, process, and record the majority of a feature-length, live-action, Hollywood film digitally. The first Hollywood film to utilize a digital intermediate process from beginning to end was O Brother, Where Art Thou? in 2000 and in Europe it was Chicken Run released that same year. The process rapidly caught on in the mid-2000s. Around 50% of Hollywood films went through a digital intermediate in 2005, increasing to around 70% by mid-2007. This is due not only to the extra creative options the process affords film makers but also the need for high-quality scanning and color adjustments to produce movies for digital cinema. == Milestones == 1990: The Rescuers Down Under – First feature-length film to be entirely recorded to film from digital files; in this case animation assembled on computers using Walt Disney Feature Animation and Pixar's CAPS system. 1992: Visual effects supervisor and producer Chris F. Woods creates a VFX studio to produce the visual effects for the 1993 film Super Mario Bros. It was the first 35mm feature film to digitally scan a large number of VFX plates (over 700) at 2K resolution, as well as to output the finished VFX to 35mm negative at 2K. 1993: Snow White and the Seven Dwarfs – First film to be entirely scanned to digital files, manipulated, and recorded back to film at 4K resolution. The restoration project was done entirely at 4K resolution and 10-bit color depth using the Cineon system to digitally remove dirt and scratches and restore faded colors. 1998: Pleasantville – The first time the majority of a new feature film was scanned, processed, and recorded digitally. The black-and-white meets color world portrayed in the movie was filmed entirely in color and selectively desaturated and contrast adjusted digitally. The work was done in Los Angeles by Cinesite utilizing a Spirit DataCine for scanning at 2K resolution and a MegaDef color correction system from UK Company Pandora International 1998: Zingo - The first feature film to use digital color correction via digital intermediate in its entirety. The work was performed at the Digital Film Lab in Copenhagen, using a Spirit Datacine to transfer the entire film to digital files at 2K resolution. The digital intermediate process was also used to perform a digital blowup of the film's original Super 16 source format to a 35mm output. 1999: Pacific Ocean Post Film, a team led by John McCunn and Greg Kimble used Kodak film scanners & laser film printer, Cineon software as well as proprietary tools to rebuild and repair the first two reels of the 1968 Beatles' film Yellow Submarine for re-release. 1999: Star Wars: Episode I – The Phantom Menace - Industrial Light & Magic (ILM) scanned the entirety of the visual effects-laden film for the purposes of digital enhancement and the integration of thousands of separately filmed elements with computer generated characters and environments. Outside of the approximately 2000 effects shots that were digitally manipulated, the remaining 170 non-effects shots were also scanned for continuity. However, after the digital shots were manipulated at ILM, they were filmed out individually and sent to Deluxe Labs where they were processed and color timed photochemically. 2000: Sorted - The first feature-length, color 35mm motion picture to fully utilize the digital intermediate process in its entirety from inception to completion. The film was produced at Wave Pictures' digital intermediate film facility in London, England. It was scanned at 2K resolution with 8 bits color depth per color / per pixel using a pin registered, liquid gate Oxberry
SitePal
SitePal is a speaking avatar platform for small and medium-sized businesses developed by Oddcast. SitePal allows users to deploy "virtual employees" on websites that can welcome visitors, guide them around the site and answer questions. The use of SitePal on commercial websites has been controversial because many visitors report finding them annoying. Some research has shown that they can increase sales in comparison to using static photographs. == Development == The technology used was the result of more than 4 years of research at Stanford University. The research was based on a literature review and other previous work in the field of artificial intelligence research. The SitePal AI option uses the AIML programming language, which is partially editable by users. This allows web designers to simulate normal human conversation by using keywords or key phrases that the bot can respond to. == Features == The company provides web designers with options to customize the chosen avatar. A large selection of faces, clothing, hair, backgrounds, voices and other details are available. If a web designer wants to use a particular face, Sitepal can create one from a photo. Thus, a mascot or a known face can be simulated. == Speech == Sitepal avatars talk through text-to-speech (tts) software. A short paragraph can be written (up to 900 characters) and the text-to-speech engine will compile the actual speech, which can be reproduced and edited. The tts engine is not perfect, but it comes close to actual speech and is easy to understand. Tts can be further enhanced by some commands, like /laugh and /loud which make the avatar laugh or talk loud. Even pronunciation is possible. The web designer can record and upload his or her own audio messages. Alternatively Sitepal offers professional voice acting service at extra cost. == User interaction == The company provides 5 options for visitor interaction: No interaction. The avatar simply says a pre-fixed message. FAQ mode. Questions can be configured, which are clickable and the user can hear the answer. Lead mode. The avatar prompts the user to type his email and short message, so it can be sent to the webmaster (usually used on a "contact us" page) Chatbot mode. The avatar greets the user, and he can type his questions and have a conversation with the bot. With predetermined replies, this can work as an FAQ as well. API customization. Experienced programmers can make their avatar interact with their website, making it talk when the user clicks on a link or when other triggers occur. Even dual avatar conversations can be created, like a talk show. == Posting options == The company provides five options for posting the avatar: Embed in webpage (via javascript) Embed in HTML Send by email Publish to eBay Embed in Flash == Criticism == Early reviews, such as one by Troy Dreier published in PC World in 2002 were positive and described SitePal as: "an engagingly simple and personal tool, and the price is reasonable for what it adds to a site". Although Dreier did note that the program had "bugs that suggested it hadn't been tested thoroughly". In more recent years, reaction to SitePal has been much more negative with reviews such as Tom Spring writing in a PC World review citing SitePal ads and described his reaction as "Not so nice". Paul Bissex, writing in E-Scribe News described SitePal as "heinous... and embarrassing if anyone is within earshot...they creep me out" == Research on effectiveness == In one single-website research project Anita Campbell had half the visitors to Small Business Trends see a SitePal and the other half see just a static photograph. Over 11,000 visitors the SitePal avatar improved sign-up for a newsletter 144% over the control condition.
Energy-based model
An energy-based model (EBM), also called Canonical Ensemble Learning (CEL) or Learning via Canonical Ensemble (LCE), is an application of canonical ensemble formulation from statistical physics for learning from data. The approach prominently appears in generative artificial intelligence. EBMs provide a unified framework for many probabilistic and non-probabilistic approaches to such learning, particularly for training graphical and other structured models. An EBM learns the characteristics of a target dataset and generates a similar but larger dataset. EBMs detect the latent variables of a dataset and generate new datasets with a similar distribution. Energy-based generative neural networks is a class of generative models, which aim to learn explicit probability distributions of data in the form of energy-based models, the energy functions of which are parameterized by modern deep neural networks. Boltzmann machines are a special form of energy-based models with a specific parametrization of the energy. == Description == For a given input x {\displaystyle x} , the model describes an energy E θ ( x ) {\displaystyle E_{\theta }(x)} such that the Boltzmann distribution P θ ( x ) = e − β E θ ( x ) Z ( θ ) {\displaystyle P_{\theta }(x)={e^{-\beta E_{\theta }(x)} \over Z(\theta )}} is a probability (density), and typically β = 1 {\displaystyle \beta =1} . Since the normalization constant: Z ( θ ) := ∫ x ∈ X e − β E θ ( x ) d x {\displaystyle Z(\theta ):=\int _{x\in X}e^{-\beta E_{\theta }(x)}dx} (also known as the partition function) depends on all the Boltzmann factors of all possible inputs x {\displaystyle x} , it cannot be easily computed or reliably estimated during training simply using standard maximum likelihood estimation. However, for maximizing the likelihood during training, the gradient of the log-likelihood of a single training example x {\displaystyle x} is given by using the chain rule: ∂ θ log ( P θ ( x ) ) = E x ′ ∼ P θ [ ∂ θ E θ ( x ′ ) ] − ∂ θ E θ ( x ) ( ∗ ) {\displaystyle \partial _{\theta }\log \left(P_{\theta }(x)\right)=\mathbb {E} _{x'\sim P_{\theta }}[\partial _{\theta }E_{\theta }(x')]-\partial _{\theta }E_{\theta }(x)\,()} The expectation in the above formula for the gradient can be approximately estimated by drawing samples x ′ {\displaystyle x'} from the distribution P θ {\displaystyle P_{\theta }} using Markov chain Monte Carlo (MCMC). Early energy-based models, such as the 2003 Boltzmann machine by Hinton, estimated this expectation via blocked Gibbs sampling. Newer approaches make use of more efficient Stochastic Gradient Langevin Dynamics (LD), drawing samples using: x 0 ′ ∼ P 0 , x i + 1 ′ = x i ′ − α 2 ∂ E θ ( x i ′ ) ∂ x i ′ + ϵ {\displaystyle x_{0}'\sim P_{0},x_{i+1}'=x_{i}'-{\frac {\alpha }{2}}{\frac {\partial E_{\theta }(x_{i}')}{\partial x_{i}'}}+\epsilon } , where ϵ ∼ N ( 0 , α ) {\displaystyle \epsilon \sim {\mathcal {N}}(0,\alpha )} . A replay buffer of past values x i ′ {\displaystyle x_{i}'} is used with LD to initialize the optimization module. The parameters θ {\displaystyle \theta } of the neural network are therefore trained in a generative manner via MCMC-based maximum likelihood estimation: the learning process follows an "analysis by synthesis" scheme, where within each learning iteration, the algorithm samples the synthesized examples from the current model by a gradient-based MCMC method (e.g., Langevin dynamics or Hybrid Monte Carlo), and then updates the parameters θ {\displaystyle \theta } based on the difference between the training examples and the synthesized ones – see equation ( ∗ ) {\displaystyle ()} . This process can be interpreted as an alternating mode seeking and mode shifting process, and also has an adversarial interpretation. Essentially, the model learns a function E θ {\displaystyle E_{\theta }} that associates low energies to correct values, and higher energies to incorrect values. After training, given a converged energy model E θ {\displaystyle E_{\theta }} , the Metropolis–Hastings algorithm can be used to draw new samples. The acceptance probability is given by: P a c c ( x i → x ∗ ) = min ( 1 , P θ ( x ∗ ) P θ ( x i ) ) . {\displaystyle P_{acc}(x_{i}\to x^{})=\min \left(1,{\frac {P_{\theta }(x^{})}{P_{\theta }(x_{i})}}\right).} == History == The term "energy-based models" was first coined in a 2003 JMLR paper where the authors defined a generalisation of independent components analysis to the overcomplete setting using EBMs. Other early work on EBMs proposed models that represented energy as a composition of latent and observable variables. == Characteristics == EBMs demonstrate useful properties: Simplicity and stability. The EBM is the only object that needs to be designed and trained. Separate networks need not be trained to ensure balance. Adaptive computation time. An EBM can generate sharp, diverse samples or (more quickly) coarse, less diverse samples. Given infinite time, this procedure produces true samples. Flexibility. In Variational Autoencoders (VAE) and flow-based models, the generator learns a map from a continuous space to a (possibly) discontinuous space containing different data modes. EBMs can learn to assign low energies to disjoint regions (multiple modes). Adaptive generation. EBM generators are implicitly defined by the probability distribution, and automatically adapt as the distribution changes (without training), allowing EBMs to address domains where generator training is impractical, as well as minimizing mode collapse and avoiding spurious modes from out-of-distribution samples. Compositionality. Individual models are unnormalized probability distributions, allowing models to be combined through product of experts or other hierarchical techniques. == Experimental results == On image datasets such as CIFAR-10 and ImageNet 32x32, an EBM model generated high-quality images relatively quickly. It supported combining features learned from one type of image for generating other types of images. It was able to generalize using out-of-distribution datasets, outperforming flow-based and autoregressive models. EBM was relatively resistant to adversarial perturbations, behaving better than models explicitly trained against them with training for classification. == Applications == Target applications include natural language processing, robotics and computer vision. The first energy-based generative neural network is the generative ConvNet proposed in 2016 for image patterns, where the neural network is a convolutional neural network. The model has been generalized to various domains to learn distributions of videos, and 3D voxels. They are made more effective in their variants. They have proven useful for data generation (e.g., image synthesis, video synthesis, 3D shape synthesis, etc.), data recovery (e.g., recovering videos with missing pixels or image frames, 3D super-resolution, etc), data reconstruction (e.g., image reconstruction and linear interpolation ). == Alternatives == EBMs compete with techniques such as variational autoencoders (VAEs), generative adversarial networks (GANs) or normalizing flows. == Extensions == === Joint energy-based models === Joint energy-based models (JEM), proposed in 2020 by Grathwohl et al., allow any classifier with softmax output to be interpreted as energy-based model. The key observation is that such a classifier is trained to predict the conditional probability p θ ( y | x ) = e f → θ ( x ) [ y ] ∑ j = 1 K e f → θ ( x ) [ j ] for y = 1 , … , K and f → θ = ( f 1 , … , f K ) ∈ R K , {\displaystyle p_{\theta }(y|x)={\frac {e^{{\vec {f}}_{\theta }(x)[y]}}{\sum _{j=1}^{K}e^{{\vec {f}}_{\theta }(x)[j]}}}\ \ {\text{ for }}y=1,\dotsc ,K{\text{ and }}{\vec {f}}_{\theta }=(f_{1},\dotsc ,f_{K})\in \mathbb {R} ^{K},} where f → θ ( x ) [ y ] {\displaystyle {\vec {f}}_{\theta }(x)[y]} is the y-th index of the logits f → {\displaystyle {\vec {f}}} corresponding to class y. Without any change to the logits it was proposed to reinterpret the logits to describe a joint probability density: p θ ( y , x ) = e f → θ ( x ) [ y ] Z ( θ ) , {\displaystyle p_{\theta }(y,x)={\frac {e^{{\vec {f}}_{\theta }(x)[y]}}{Z(\theta )}},} with unknown partition function Z ( θ ) {\displaystyle Z(\theta )} and energy E θ ( x , y ) = − f θ ( x ) [ y ] {\displaystyle E_{\theta }(x,y)=-f_{\theta }(x)[y]} . By marginalization, we obtain the unnormalized density p θ ( x ) = ∑ y p θ ( y , x ) = ∑ y e f → θ ( x ) [ y ] Z ( θ ) =: e − E θ ( x ) , {\displaystyle p_{\theta }(x)=\sum _{y}p_{\theta }(y,x)=\sum _{y}{\frac {e^{{\vec {f}}_{\theta }(x)[y]}}{Z(\theta )}}=:e^{-E_{\theta }(x)},} therefore, E θ ( x ) = − log ( ∑ y e f → θ ( x ) [ y ] Z ( θ ) ) , {\displaystyle E_{\theta }(x)=-\log \left(\sum _{y}{\frac {e^{{\vec {f}}_{\theta }(x)[y]}}{Z(\theta )}}\right),} so that any classifier can be used to define an energy function E θ ( x ) {\displaystyle E_{\theta }(x)} .
Power cycling
Power cycling is the act of turning a piece of equipment, usually a computer, off and then on again. Reasons for power cycling include having an electronic device reinitialize its set of configuration parameters or recover from an unresponsive state of its mission critical functionality, such as in a crash or hang situation. Power cycling can also be used to reset network activity inside a modem. It can also be among the first steps for troubleshooting an issue. == Overview == Power cycling can be done manually, usually using the power switch on the device, or remotely, through some type of external device connected to the power input. In the data center environment, remote control power cycling can usually be done through a power distribution unit, over the network. In the home environment, this can be done through home automation powerline communications. Most Internet service providers publish a "how-to" on their website showing their customers the correct procedure to power cycle their devices. Power cycling is a common diagnostic procedure usually performed first when a computer system freezes. However, frequently power cycling a computer can cause thermal stress. Reset has an equal effect on the software but may be less problematic for the hardware as power is not interrupted. == Historical uses == On all Apollo missions to the moon, the landing radar was required to acquire the surface before a landing could be attempted. But on Apollo 14, the landing radar was unable to lock on. Mission control told the astronauts to cycle the power. They did, the radar locked on just in time, and the landing was completed. During the Rosetta mission to comet 67P/Churyumov–Gerasimenko, the Philae lander did not return the expected telemetry on awakening after arrival at the comet. The problem was diagnosed as "somehow a glitch in the electronics", engineers cycled the power, and the lander awoke correctly. During the launch of the billion dollar AEHF-6 satellite on 26 March 2020 by an Atlas V rocket from Cape Canaveral Space Force Station in Florida, a hold was called at T-46 seconds due to hydraulic system not responding as expected. The launch crew turned it off and back on, and the launch proceeded normally. In 2023 the Interstellar Boundary Explorer spacecraft stopped responding to commands after an anomaly. When gentler techniques failed, NASA resorted to rebooting the spacecraft with the remote equivalent of a power cycle.