Jubatus is an open-source online machine learning and distributed computing framework developed at Nippon Telegraph and Telephone and Preferred Infrastructure. Its features include classification, recommendation, regression, anomaly detection and graph mining. It supports many client languages, including C++, Java, Ruby and Python. It uses Iterative Parameter Mixture for distributed machine learning. == Notable Features == Jubatus supports: Multi-classification algorithms: Perceptron Passive Aggressive Confidence Weighted Adaptive Regularization of Weight Vectors Normal Herd Recommendation algorithms using: Inverted index Minhash Locality-sensitive hashing Regression algorithms: Passive Aggressive feature extraction method for natural language: n-gram Text segmentation
Super-resolution optical fluctuation imaging
Super-resolution optical fluctuation imaging (SOFI) is a post-processing method for the calculation of super-resolved images from recorded image time series that is based on the temporal correlations of independently fluctuating fluorescent emitters. SOFI has been developed for super-resolution of biological specimen that are labelled with independently fluctuating fluorescent emitters (organic dyes, fluorescent proteins). In comparison to other super-resolution microscopy techniques such as STORM or PALM that rely on single-molecule localization and hence only allow one active molecule per diffraction-limited area (DLA) and timepoint, SOFI does not necessitate a controlled photoswitching and/ or photoactivation as well as long imaging times. Nevertheless, it still requires fluorophores that are cycling through two distinguishable states, either real on-/off-states or states with different fluorescence intensities. In mathematical terms SOFI-imaging relies on the calculation of cumulants, for what two distinguishable ways exist. For one thing an image can be calculated via auto-cumulants that by definition only rely on the information of each pixel itself, and for another thing an improved method utilizes the information of different pixels via the calculation of cross-cumulants. Both methods can increase the final image resolution significantly although the cumulant calculation has its limitations. Actually SOFI is able to increase the resolution in all three dimensions. == Principle == Likewise to other super-resolution methods SOFI is based on recording an image time series on a CCD- or CMOS camera. In contrary to other methods the recorded time series can be substantially shorter, since a precise localization of emitters is not required and therefore a larger quantity of activated fluorophores per diffraction-limited area is allowed. The pixel values of a SOFI-image of the n-th order are calculated from the values of the pixel time series in the form of a n-th order cumulant, whereas the final value assigned to a pixel can be imagined as the integral over a correlation function. The finally assigned pixel value intensities are a measure of the brightness and correlation of the fluorescence signal. Mathematically, the n-th order cumulant is related to the n-th order correlation function, but exhibits some advantages concerning the resulting resolution of the image. Since in SOFI several emitters per DLA are allowed, the photon count at each pixel results from the superposition of the signals of all activated nearby emitters. The cumulant calculation now filters the signal and leaves only highly correlated fluctuations. This provides a contrast enhancement and therefore a background reduction for good measure. As it is implied in the figure on the left the fluorescence source distribution: ∑ k = 1 N δ ( r → − r → k ) ⋅ ε k ⋅ s k ( t ) {\displaystyle \sum _{k=1}^{N}\delta ({\vec {r}}-{\vec {r}}_{k})\cdot \varepsilon _{k}\cdot s_{k}(t)} is convolved with the system's point spread function (PSF) U(r). Hence the fluorescence signal at time t and position r → {\displaystyle {\vec {r}}} is given by F ( r → , t ) = ∑ k = 1 N U ( r → − r → k ) ⋅ ε k ⋅ s k ( t ) . {\displaystyle F({\vec {r}},t)=\sum _{k=1}^{N}U({\vec {r}}-{\vec {r}}_{k})\cdot \varepsilon _{k}\cdot s_{k}(t).} Within the above equations N is the amount of emitters, located at the positions r → k {\displaystyle {\vec {r}}_{k}} with a time-dependent molecular brightness ε k ⋅ s k {\displaystyle \varepsilon _{k}\cdot s_{k}} where ε k {\displaystyle \varepsilon _{k}} is a variable for the constant molecular brightness and s k ( t ) {\displaystyle s_{k}(t)} is a time-dependent fluctuation function. The molecular brightness is just the average fluorescence count-rate divided by the number of molecules within a specific region. For simplification it has to be assumed that the sample is in a stationary equilibrium and therefore the fluorescence signal can be expressed as a zero-mean fluctuation: δ F ( r → , t ) = F ( r → , t ) − ⟨ F ( r → , t ) ⟩ t {\displaystyle \delta F({\vec {r}},t)=F({\vec {r}},t)-\langle F({\vec {r}},t)\rangle _{t}} where ⟨ ⋯ ⟩ t {\displaystyle \langle \cdots \rangle _{t}} denotes time-averaging. The auto-correlation here e.g. the second-order can then be described deductively as follows for a certain time-lag τ {\displaystyle \tau } : δ F ( r → , t ) = ⟨ δ F ( r → , t + τ ) ⋅ δ F ( r → , t ) ⟩ t {\displaystyle \delta F({\vec {r}},t)=\langle \delta F({\vec {r}},t+\tau )\cdot \delta F({\vec {r}},t)\rangle _{t}} From these equations it follows that the PSF of the optical system has to be taken to the power of the order of the correlation. Thus in a second-order correlation the PSF would be reduced along all dimensions by a factor of 2 {\displaystyle {\sqrt {2}}} . As a result, the resolution of the SOFI-images increases according to this factor. === Cumulants versus correlations === Using only the simple correlation function for a reassignment of pixel values, would ascribe to the independency of fluctuations of the emitters in time in a way that no cross-correlation terms would contribute to the new pixel value. Calculations of higher-order correlation functions would suffer from lower-order correlations for what reason it is superior to calculate cumulants, since all lower-order correlation terms vanish. == Cumulant-calculation == === Auto-cumulants === For computational reasons it is convenient to set all time-lags in higher-order cumulants to zero so that a general expression for the n-th order auto-cumulant can be found: A C n ( r → , τ 1 … n − 1 = 0 ) = ∑ k = 1 N U n ( r → − r → k ) ε k n w k ( 0 ) {\displaystyle AC_{n}({\vec {r}},\tau _{1\ldots n-1}=0)=\sum _{k=1}^{N}U^{n}({\vec {r}}-{\vec {r}}_{k})\varepsilon _{k}^{n}w_{k}(0)} w k {\displaystyle w_{k}} is a specific correlation based weighting function influenced by the order of the cumulant and mainly depending on the fluctuation properties of the emitters. Albeit there is no fundamental limitation in calculating very high orders of cumulants and thereby shrinking the FWHM of the PSF there are practical limitations according to the weighting of the values assigned to the final image. Emitters with a higher molecular brightness will show a strong increase in terms of the pixel cumulant value assigned at higher-orders as well as this performance can be expected from a diverse appearance of fluctuations of different emitters. A wide intensity range of the resulting image can therefore be expected and as a result dim emitters can get masked by bright emitters in higher-order images:. The calculation of auto-cumulants can be realized in a very attractive way in a mathematical sense. The n-th order cumulant can be calculated with a basic recursion from moments K n ( r → ) = μ n ( r → ) − ∑ i = 1 n − 1 ( n − 1 i ) K n − i ( r → ) μ i ( r → ) {\displaystyle K_{n}({\vec {r}})=\mu _{n}({\vec {r}})-\sum _{i=1}^{n-1}{\begin{pmatrix}n-1\\i\end{pmatrix}}K_{n-i}({\vec {r}})\mu _{i}({\vec {r}})} where K is a cumulant of the index's order, likewise μ {\displaystyle \mu } represents the moments. The term within the brackets indicates a binomial coefficient. This way of computation is straightforward in comparison with calculating cumulants with standard formulas. It allows for the calculation of cumulants with only little time of computing and is, as it is well implemented, even suitable for the calculation of high-order cumulants on large images. === Cross-cumulants === In a more advanced approach cross-cumulants are calculated by taking the information of several pixels into account. Cross-cumulants can be described as follows: C C n ( r → , τ 1 … n − 1 = 0 ) = ∏ j < l n U ( r → j − r → l n ) ⋅ ∑ i = 1 N U n ( r → i − ∑ k n r → k n ) ε i n w i ( 0 ) {\displaystyle CC_{n}({\vec {r}},\tau _{1\ldots n-1}=0)=\prod _{j A SIGINT Activity Designator (or SIGAD) identifies a signals intelligence (SIGINT) line of collection activity associated with a signals collection station, such as a base or a ship. For example, the SIGAD for Menwith Hill in the UK is USD1000. SIGADs are used by the signals intelligence agencies of Australia, Canada, New Zealand, the United Kingdom, and the United States (the Five Eyes). There are several thousand SIGADs including the substation SIGADs denoted with a trailing alpha character. Several dozen of these are significant. The leaked Boundless Informant reporting screenshot showed that it summarized 504 active SIGADs during a 30-day period in March 2013. == General format == A SIGAD consists of five to eight case insensitive alphanumeric characters. It takes the general form of an alphanumeric designator normally composed of a two- or three-letter prefix followed by one to three numbers. Often a dash is used to separate the alphabetic and numeric characters in the primary part of the designator, but less frequently a space is used as a separator or the alphabetic and numeric characters are concatenated together. An additional alphabetic character can be added to denote a sub-designator for a subset of the primary unit, such as a detachment. Lastly, a numeric character can be added after the aforementioned alphabetic to provide for a sub-sub-designator. In the examples below an X represents an alphabetic character and an N represents a numeric character that are part of the primary designator. Likewise, an x represents an alphabetic character and an n represents a numeric character that are part of a sub-designator. Here are valid generalized examples of SIGADs: The first two characters show which country operates the particular SIGINT facility, which can be US for the United States, UK for the United Kingdom, CA for Canada, AU for Australia and NZ for New Zealand. A third letter shows what sort of staff runs the station. SIGADs beginning with US without a third letter are used for intercept facilities run by the NSA. == PRISM SIGAD == One prominent SIGAD as of April 2013 is US-984XN, with an unclassified codename of PRISM. It is "the number one source of raw intelligence used for NSA analytic reports" according to National Security Agency sources in a document leaked by Edward Snowden. The President's Daily Brief, an all-source intelligence product, cited SIGAD US-984XN as a source in 1,477 items in 2012. The U.S. government operates the PRISM electronic surveillance collection program through NSA's Special Source Operations, an alliance with trusted telecommunications providers. == SIGADs for spy ships == The declassified SIGAD for the USS Liberty (AGTR-5) was USN-855. The USS Liberty incident occurred on 8 June 1967, during the Six-Day War, when Israeli Air Force jet fighter aircraft and Israeli Navy motor torpedo boats attacked the USS Liberty in international waters. The USS Pueblo (AGER-2) was a technical research ship, which was boarded and captured by North Korean forces on 23 January 1968, in what is known as the Pueblo incident. The declassified SIGAD for the NSA Direct Support Unit (DSU) from the Naval Security Group (NSG) on the USS Pueblo patrol involved in the incident was USN-467Y. The USS Pueblo, which officially remains a commissioned vessel of the United States Navy, is the only ship of the U.S. Navy currently being held captive. == Vietnam War SIGADs == The following are the Vietnam War-era declassified SIGADs from inside South Vietnam during the period of 1969 to 1975: Some locations have multiple SIGADs due to different types of collection activities and/or collection at different times during the period. The SIGADs beginning with USA were operated by the United States Air Force's United States Air Force Security Service (USAFSS). The SIGADs beginning with USM were operated by the United States Army's Army Security Agency (ASA). Lastly, the SIGADs beginning with USN were operated by the United States Navy's Naval Security Group (NAVSECGRU). All three of these units have been merged into other units or inactivated. The above list consists of the higher-echelon SIGADs. It does not include the numerous miscellaneous and temporary detachments, or direction finding stations belonging to major units or sites unless that detachment or site was the only one stationed in South Vietnam. Many of the "dets" were short-lived, often formed to support ongoing MACV operations or forward deployments of combat operational or maneuver units. These detachments usually were designated by a letter suffix attached to the higher-echelon SIGAD such as "USM-633J," which was a detachment of the 372d Radio Research Company, USM-633, supporting the United States Army's 25th Infantry Division. === Supporting Southeast Asia SIGADs === The following declassified SIGADs were highly relevant to the Vietnam Campaign, but were located in areas outside of South Vietnam in Southeast Asia. Again, detachments are not listed separately. In the case of the USS Maddox, naval Direct Support Units (DSUs) used the SIGAD USN-467 as a generic designator for their missions. Each specific patrol received a letter suffix for its duration. The subsequent mission would receive the next letter in an alphabetic sequence. Thus, SIGAD USN-467N specifically designates the USS Maddox patrol involved with the Gulf of Tonkin incident. == Joint Base SIGADs == In November 2005, the US Congress performed a fifth round of Base Realignment and Closure. This 2005 law also created twelve joint bases by merging adjacent installations belonging to different services in an effort to reduce costs and improve efficiencies. Joint bases with a primarily SIGINT mission have SIGADs that begin with USJ. A joint base would have a primary SIGAD in the general form of USJ-NNN, where NNN are numeric characters. An actual example is not given, since these units are currently active. Developed at UC Berkeley, "Opinion Space" (also known as The Collective Discovery Engine) is a social media technology designed to help communities generate and exchange ideas about important issues and policies. Version 1.0 was launched on April 4, 2009, at UC Berkeley, and explored the question "Do you think legalizing marijuana is a good idea?" It has since undergone 4 different iterations, and been used in partnership with various organizations including The Occupy movement (Version 4.0, 5/24/2013) and the African Robots Network (Version 4.0, 5/25/2013). Opinion Space has also been used in collaboration with the United States State Department and the University of California's Berkeley Center for New Media (Version 2.0, 12/1/2009 and Version 3.0, 2/25/2012) to gain public perspective on foreign policy issues. Then U.S. Secretary of State Hillary Rodham Clinton explained, "Opinion Space will harness the power of connection technologies to provide a unique forum for international dialogue. This is...an opportunity to extend our engagement beyond the halls of government directly to the people of the world" (2010). The website uses data visualization and statistical analysis to present and develop public opinion and ideas. Opinion Space is a self-organizing system that uses an intuitive graphical "map" that displays patterns, trends, and insights as they emerge and employs the wisdom of crowds to identify and highlight the most insightful ideas. The system uses a game model that incorporates techniques from deliberative polling, collaborative filtering, and multidimensional visualization. In cryptography and the theory of computation, Yao's test is a test defined by Andrew Chi-Chih Yao in 1982, against pseudo-random sequences. A sequence of words passes Yao's test if an attacker with reasonable computational power cannot distinguish it from a sequence generated uniformly at random. == Formal statement == === Boolean circuits === Let P {\displaystyle P} be a polynomial, and S = { S k } k {\displaystyle S=\{S_{k}\}_{k}} be a collection of sets S k {\displaystyle S_{k}} of P ( k ) {\displaystyle P(k)} -bit long sequences, and for each k {\displaystyle k} , let μ k {\displaystyle \mu _{k}} be a probability distribution on S k {\displaystyle S_{k}} , and P C {\displaystyle P_{C}} be a polynomial. A predicting collection C = { C k } {\displaystyle C=\{C_{k}\}} is a collection of boolean circuits of size less than P C ( k ) {\displaystyle P_{C}(k)} . Let p k , S C {\displaystyle p_{k,S}^{C}} be the probability that on input s {\displaystyle s} , a string randomly selected in S k {\displaystyle S_{k}} with probability μ ( s ) {\displaystyle \mu (s)} , C k ( s ) = 1 {\displaystyle C_{k}(s)=1} , i.e. Moreover, let p k , U C {\displaystyle p_{k,U}^{C}} be the probability that C k ( s ) = 1 {\displaystyle C_{k}(s)=1} on input s {\displaystyle s} a P ( k ) {\displaystyle P(k)} -bit long sequence selected uniformly at random in { 0 , 1 } P ( k ) {\displaystyle \{0,1\}^{P(k)}} . We say that S {\displaystyle S} passes Yao's test if for all predicting collection C {\displaystyle C} , for all but finitely many k {\displaystyle k} , for all polynomial Q {\displaystyle Q} : === Probabilistic formulation === As in the case of the next-bit test, the predicting collection used in the above definition can be replaced by a probabilistic Turing machine, working in polynomial time. This also yields a strictly stronger definition of Yao's test (see Adleman's theorem). Indeed, one could decide undecidable properties of the pseudo-random sequence with the non-uniform circuits described above, whereas BPP machines can always be simulated by exponential-time deterministic Turing machines. Label noise refers to errors or inaccuracies in the class labels of data instances. This is a widespread issue in machine learning datasets, arising from human annotator mistakes, unclear labeling instructions, automated labeling methods, or adversarial attacks in supervised learning. Label noise can be roughly divided into random noise, where labels are flipped independently of input features, and systematic noise, where mislabeling is dependent on certain patterns or biases in the data. Label noise can be damaging to model performance, especially for complex models that may overfit to noisy labels rather than generalizable patterns. Many approaches have been proposed to deal with the effects of label noise, including robust loss functions, noise-tolerant algorithms, data cleaning methods, and semi-supervised learning approaches. To reduce the impact of wrong labels during training, techniques like label smoothing, sample reweighting and using trusted validation sets are used. The role of noise-robust training paradigms and curriculum learning strategies to improve resilience against mislabeled data is also explored in recent research. Cut, copy, and paste are essential commands of modern human–computer interaction and user interface design. They offer an interprocess communication technique for transferring data through a computer's user interface. The cut command removes the selected data from its original position, and the copy command creates a duplicate; in both cases the selected data is kept in temporary storage called the clipboard. Clipboard data is later inserted wherever a paste command is issued. The data remains available to any application supporting the feature, thus allowing easy data transfer between applications. The command names are a (skeuomorphic) interface metaphor based on the physical procedure used in manuscript print editing to create a page layout, like with paper. The commands were pioneered into computing by Xerox PARC in 1974, popularized by Apple Computer in the 1983 Lisa workstation and the 1984 Macintosh computer, and in a few home computer applications such as the 1984 word processor Cut & Paste. This interaction technique has close associations with related techniques in graphical user interfaces (GUIs) that use pointing devices such as a computer mouse (by drag and drop, for example). Typically, clipboard support is provided by an operating system as part of its GUI and widget toolkit. The capability to replicate information with ease, changing it between contexts and applications, involves privacy concerns because of the risks of disclosure when handling sensitive information. Terms like cloning, copy forward, carry forward, or re-use refer to the dissemination of such information through documents, and may be subject to regulation by administrative bodies. == History == === Origins === The term "cut and paste" comes from the traditional practice in manuscript editing, whereby people cut paragraphs from a page with scissors and paste them onto another page. This practice remained standard into the 1980s. Stationery stores sold "editing scissors" with blades long enough to cut an 8½"-wide page. The advent of photocopiers made the practice easier and more flexible. The act of copying or transferring text from one part of a computer-based document ("buffer") to a different location within the same or different computer-based document was a part of the earliest on-line computer editors. As soon as computer data entry moved from punch-cards to online files (in the mid/late 1960s) there were "commands" for accomplishing this operation. This mechanism was often used to transfer frequently-used commands or text snippets from additional buffers into the document, as was the case with the QED text editor. === Early methods === The earliest editors (designed for teleprinter terminals) provided keyboard commands to delineate a contiguous region of text, then delete or move it. Since moving a region of text requires first removing it from its initial location and then inserting it into its new location, various schemes had to be invented to allow for this multi-step process to be specified by the user. Often this was done with a "move" command, but some text editors required that the text be first put into some temporary location for later retrieval/placement. In 1983, the Apple Lisa became the first text editing system to call that temporary location "the clipboard". Earlier control schemes such as NLS used a verb—object command structure, where the command name was provided first and the object to be copied or moved was second. The inversion from verb—object to object—verb on which copy and paste are based, where the user selects the object to be operated before initiating the operation, was an innovation crucial for the success of the desktop metaphor as it allowed copy and move operations based on direct manipulation. === Popularization === Inspired by early line and character editors, such as Pentti Kanerva's TV-Edit, that broke a move or copy operation into two steps—between which the user could invoke a preparatory action such as navigation—Lawrence G. "Larry" Tesler proposed the names "cut" and "copy" for the first step and "paste" for the second step. Beginning in 1974, he and colleagues at Xerox PARC implemented several text editors that used cut/copy-and-paste commands to move and copy text. Apple Computer popularized this paradigm with its Lisa (1983) and Macintosh (1984) operating systems and applications. The functions were mapped to key combinations using the ⌘ Command key as a special modifier, which is held down while also pressing X for cut, C for copy, or V for paste. These few keyboard shortcuts allow the user to perform all the basic editing operations, and the keys are clustered at the left end of the bottom row of the standard QWERTY keyboard. These are the standard shortcuts: Control-Z (or ⌘ Command+Z) to undo Control-X (or ⌘ Command+X) to cut Control-C (or ⌘ Command+C) to copy Control-V (or ⌘ Command+V) to paste The IBM Common User Access (CUA) standard also uses combinations of the Insert, Del, Shift and Control keys. Early versions of Windows used the IBM standard. Microsoft later also adopted the Apple key combinations with the introduction of Windows, using the control key as modifier key. Similar patterns of key combinations, later borrowed by others, are widely available in most GUI applications. The original cut, copy, and paste workflow, as implemented at PARC, utilizes a unique workflow: With two windows on the same screen, the user could use the mouse to pick a point at which to make an insertion in one window (or a segment of text to replace). Then, by holding shift and selecting the copy source elsewhere on the same screen, the copy would be made as soon as the shift was released. Similarly, holding shift and control would copy and cut (delete) the source. This workflow requires many fewer keystrokes/mouse clicks than the current multi-step workflows, and did not require an explicit copy buffer. It was dropped, one presumes, because the original Apple and IBM GUIs were not high enough density to permit multiple windows, as were the PARC machines, and so multiple simultaneous windows were rarely used. == Cut and paste == Computer-based editing can involve very frequent use of cut-and-paste operations. Most software-suppliers provide several methods for performing such tasks, and this can involve (for example) key combinations, pulldown menus, pop-up menus, or toolbar buttons. The user selects or "highlights" the text or file for moving by some method, typically by dragging over the text or file name with the pointing-device or holding down the Shift key while using the arrow keys to move the text cursor. The user performs a "cut" operation via key combination Ctrl+x (⌘+x for Macintosh users), menu, or other means. Visibly, "cut" text immediately disappears from its location. "Cut" files typically change color to indicate that they will be moved. Conceptually, the text has now moved to a location often called the clipboard. The clipboard typically remains invisible. On most systems only one clipboard location exists, hence another cut or copy operation overwrites the previously stored information. Many UNIX text-editors provide multiple clipboard entries, as do some Macintosh programs such as Clipboard Master, and Windows clipboard-manager programs such as the one in Microsoft Office. The user selects a location for insertion by some method, typically by clicking at the desired insertion point. A paste operation takes place which visibly inserts the clipboard text at the insertion point. (The paste operation does not typically destroy the clipboard text: it remains available in the clipboard and the user can insert additional copies at other points). Whereas cut-and-paste often takes place with a mouse-equivalent in Windows-like GUI environments, it may also occur entirely from the keyboard, especially in UNIX text editors, such as Pico or vi. Cutting and pasting without a mouse can involve a selection (for which Ctrl+x is pressed in most graphical systems) or the entire current line, but it may also involve text after the cursor until the end of the line and other more sophisticated operations. The clipboard usually stays invisible, because the operations of cutting and pasting, while actually independent, usually take place in quick succession, and the user (usually) needs no assistance in understanding the operation or maintaining mental context. Some application programs provide a means of viewing, or sometimes even editing, the data on the clipboard. == Copy and paste == The term "copy-and-paste" refers to the popular, simple method of reproducing text or other data from a source to a destination. It differs from cut and paste in that the original source text or data does not get deleted or removed. The popularity of this method stems from its simplicity and the ease with which users can move data between various applications visually – without resorting to permanent storage. Use in healthcare doSIGINT Activity Designator
Opinion Space
Yao's test
Label noise
Cut, copy, and paste