In computing, a hardware random number generator (HRNG), true random number generator (TRNG), non-deterministic random bit generator (NRBG), or physical random number generator is a device that generates random numbers from a physical process capable of producing entropy, unlike a pseudorandom number generator (PRNG) that utilizes a deterministic algorithm and non-physical nondeterministic random bit generators that do not include hardware dedicated to generation of entropy. Many natural phenomena generate low-level, statistically random "noise" signals, including thermal and shot noise, jitter and metastability of electronic circuits, Brownian motion, and atmospheric noise. Researchers also used the photoelectric effect, involving a beam splitter, other quantum phenomena, and even nuclear decay (due to practical considerations the latter, as well as the atmospheric noise, is not viable except for fairly restricted applications or online distribution services). While "classical" (non-quantum) phenomena are not truly random, an unpredictable physical system is usually acceptable as a source of randomness, so the qualifiers "true" and "physical" are used interchangeably. A hardware random number generator is expected to output near-perfect random numbers ("full entropy"). A physical process usually does not have this property, and a practical TRNG typically includes a few blocks: a noise source that implements the physical process producing the entropy. Usually this process is analog, so a digitizer is used to convert the output of the analog source into a binary representation; a conditioner (randomness extractor) that improves the quality of the random bits; health tests. TRNGs are mostly used in cryptographical algorithms that get completely broken if the random numbers have low entropy, so the testing functionality is usually included. Hardware random number generators generally produce only a limited number of random bits per second. In order to increase the available output data rate, they are often used to generate the "seed" for a faster PRNG. PRNG also helps with the noise source "anonymization" (whitening out the noise source identifying characteristics) and entropy extraction. With a proper PRNG algorithm selected (cryptographically secure pseudorandom number generator, CSPRNG), the combination can satisfy the requirements of Federal Information Processing Standards and Common Criteria standards. == Uses == Hardware random number generators can be used in any application that needs randomness. However, in many scientific applications additional cost and complexity of a TRNG (when compared with pseudo random number generators) provide no meaningful benefits. TRNGs have additional drawbacks for data science and statistical applications: impossibility to re-run a series of numbers unless they are stored, reliance on an analog physical entity can obscure the failure of the source. The TRNGs therefore are primarily used in the applications where their unpredictability and the impossibility to re-run the sequence of numbers are crucial to the success of the implementation: in cryptography and gambling machines. === Cryptography === The major use for hardware random number generators is in the field of data encryption, for example to create random cryptographic keys and nonces needed to encrypt and sign data. In addition to randomness, there are at least two additional requirements imposed by the cryptographic applications: forward secrecy guarantees that the knowledge of the past output and internal state of the device should not enable the attacker to predict future data; backward secrecy protects the "opposite direction": knowledge of the output and internal state in the future should not divulge the preceding data. A typical way to fulfill these requirements is to use a TRNG to seed a cryptographically secure pseudorandom number generator. == History == Physical devices were used to generate random numbers for thousands of years, primarily for gambling. Dice in particular have been known for more than 5000 years (found on locations in modern Iraq and Iran), and flipping a coin (thus producing a random bit) dates at least to the times of ancient Rome. The first documented use of a physical random number generator for scientific purposes was by Francis Galton (1890). He devised a way to sample a probability distribution using a common gambling die. In addition to the top digit, Galton also looked at the face of a die closest to him, thus creating 64 = 24 outcomes (about 4.6 bits of randomness). Kendall and Babington-Smith (1938) used a fast-rotating 10-sector disk that was illuminated by periodic bursts of light. The sampling was done by a human who wrote the number under the light beam onto a pad. The device was utilized to produce a 100,000-digit random number table (at the time such tables were used for statistical experiments, like PRNG nowadays). On 29 April 1947, the RAND Corporation began generating random digits with an "electronic roulette wheel", consisting of a random frequency pulse source of about 100,000 pulses per second gated once per second with a constant frequency pulse and fed into a five-bit binary counter. Douglas Aircraft built the equipment, implementing Cecil Hasting's suggestion (RAND P-113) for a noise source (most likely the well known behavior of the 6D4 miniature gas thyratron tube, when placed in a magnetic field). Twenty of the 32 possible counter values were mapped onto the 10 decimal digits and the other 12 counter values were discarded. The results of a long run from the RAND machine, filtered and tested, were converted into a table, which originally existed only as a deck of punched cards, but was later published in 1955 as a book, 50 rows of 50 digits on each page (A Million Random Digits with 100,000 Normal Deviates). The RAND table was a significant breakthrough in delivering random numbers because such a large and carefully prepared table had never before been available. It has been a useful source for simulations, modeling, and for deriving the arbitrary constants in cryptographic algorithms to demonstrate that the constants had not been selected maliciously ("nothing up my sleeve numbers"). Since the early 1950s, research into TRNGs has been highly active, with thousands of research works published and about 2000 patents granted by 2017. == Physical phenomena with random properties == Multiple different TRNG designs were proposed over time with a large variety of noise sources and digitization techniques ("harvesting"). However, practical considerations (size, power, cost, performance, robustness) dictate the following desirable traits: use of a commonly available inexpensive silicon process; exclusive use of digital design techniques. This allows an easier system-on-chip integration and enables the use of FPGAs; compact and low-power design. This discourages use of analog components (e.g., amplifiers); mathematical justification of the entropy collection mechanisms. Stipčević & Koç in 2014 classified the physical phenomena used to implement TRNG into four groups: electrical noise; free-running oscillators; chaos; quantum effects. === Electrical noise-based RNG === Noise-based RNGs generally follow the same outline: the source of a noise generator is fed into a comparator. If the voltage is above threshold, the comparator output is 1, otherwise 0. The random bit value is latched using a flip-flop. Sources of noise vary and include: Johnson–Nyquist noise ("thermal noise"); Zener noise; avalanche breakdown. The drawbacks of using noise sources for an RNG design are: noise levels are hard to control, they vary with environmental changes and device-to-device; calibration processes needed to ensure a guaranteed amount of entropy are time-consuming; noise levels are typically low, thus the design requires power-hungry amplifiers. The sensitivity of amplifier inputs enables manipulation by an attacker; circuitry located nearby generates a lot of non-random noise thus lowering the entropy; a proof of randomness is near-impossible as multiple interacting physical processes are involved. === Chaos-based RNG === The idea of chaos-based noise stems from the use of a complex system that is hard to characterize by observing its behavior over time. For example, lasers can be put into (undesirable in other applications) chaos mode with chaotically fluctuating power, with power detected using a photodiode and sampled by a comparator. The design can be quite small, as all photonics elements can be integrated on-chip. Stipčević & Koç characterize this technique as "most objectionable", mostly due to the fact that chaotic behavior is usually controlled by a differential equation and no new randomness is introduced, thus there is a possibility of the chaos-based TRNG producing a limited subset of possible output strings. === Free-running oscillators-based RNG === The TRNGs based on a free-running oscilla
Friendica
Friendica (formerly Friendika, originally Mistpark) is a free and open-source software distributed social network. It forms one part of the Fediverse, an interconnected and decentralized network of independently operated servers. == Features == Friendica users can connect with others via their own Friendica server, but may also fully integrate contacts from other platforms including Diaspora, Pump.io, GNU social, email, Discourse and more recently ActivityPub (including Mastodon, Pleroma and Pixelfed) and Bluesky into their 'newsfeed'. In addition to these two way connections, users can also use Friendica as a publishing platform to post content to WordPress, Tumblr, Insanejournal and Libertree. Posting to Google+ was also supported until that service was shut down. In addition, RSS feeds can be ingested. Because users are distributed across many servers, their "addresses" consist of a username, the "@" symbol, and the domain name of the Friendica instance in the same manner email addresses are formed. Twitter support was available but was deprecated due to API changes under Elon Musk's leadership rendering it unusable. Most of the functionality from major microblogging and social networking platforms are available in Friendica; for example, tagging users and groups via "@ mentions"; direct messages; hashtags; photo albums; "likes"; "dislikes"; comments; and re-shares of publicly visible posts. Published items can be edited and updated across the network. Comprehensive settings for privacy and the public visibility of posts allow users to regulate who can read which contributions, or see specific information about the user. Users can also create multiple profiles, allowing different groups of people (such as friends, or work mates) to see a different profile entirely when viewing the same page. User accounts can be downloaded or deleted, and can be imported to a different Friendica server if so required. Public forums can be created under different accounts, which can be switched between if the accounts are registered with the same email address. == Development == There is no corporation behind Friendica. The developers work on a voluntary basis and the project is run informally; the platform itself is used for the communication between the developers. There are different forums within Friendica, such as "Friendica Developers" and "Friendica Support". The source code of Friendica is hosted on GitHub. == Installation == The developers aim to make installation of the software as simple as possible for technical laymen. They argue that decentralization on small servers is a key condition for the freedom of users and their self-determination. The difficulty level is similar to an installation of WordPress. However, the installing on shared hosting is sometimes difficult because of missing PHP5 modules. Some volunteers also run public servers so that newcomers can also avoid the installation of their own software. == List of clients == Friendica implements multiple client-server API variants simultaneously. Along with endpoints needed to use enhanced Friendica features, it also implements the API used by GNU social, Twitter and since version 2021.06 also the one used by Mastodon. As a result, most GNU social and Mastodon clients can be used for Friendica. Examples of Friendica compatible clients include: Raccoon for Friendica, Friendiqa, Fedilab, AndStatus, Twidere and DiCa for Android, friendly for Sailfish OS, friclicli (CLI client), choqok and Friendiqa for Linux and Friendica Mobile for Windows 10. == Reception == Friendica was cited in January 2012 by Infoshop News as an "alternative to Google+ and Facebook" to be used on the Occupy Nigeria movement. In January 2012 Free Software Foundation Europe's blog cited Friendica as a reasonable alternative to centralized and controlled social networks such as Facebook or Google+. Biblical Notes writer J. Randal Matheny described Friendica in January 2012 as "One social networking option flying under the radar until recently deserves consideration as an already stable platform with a wide range of options, applications, plug-ins, and possibilities for opening up the Internet." In February 2012, the German computer magazine c't wrote: "Friendica demonstrates how decentralized social networks can become widely accepted." Another German publication, the professional magazine t3n listed Friendica as a Facebook rival in an online article in March 2012 about Facebook alternatives. It compared Friendica with similar social networks like Diaspora and identi.ca. MSN Tech & Gadgets contributor Emma Boyes wrote about Friendica in May 2012: "why you'll love it: you can use it to access all the other social networks and get recommendations of new friends and groups to join. Friendica is open source and decentralised. There's no corporation behind it and there are extensive privacy settings. You can choose from a variety of user interfaces and it boasts some cool features—for instance, being able to key in a list of your interests and use the 'profile match' feature to recommend other users who share them with you. A word of warning, though, the site is not as user-friendly as the others on this list, so it may be this one is one for the geeks." == Later reviews == Acquisition of Twitter by Elon Musk had revitalized public interest in Fediverse technologies in April 2022. Friendica received favorable reviews, with a PCMag article describing it as "mostly comparable to Facebook", drawing a parallel to Google+ and highlighting using it "for planning events, and its multiple profile feature means you can show a different face to your friends, coworkers, and family". The September 2022 issue of Linux Magazine contains a detailed comparison and walk-through of registering to and using basic functions of Diaspora, Friendica and Mastodon. They describe Friendica as "intuitive" and highlight the "huge choice of account settings" and that "Friendica does not require any specific hardware, so you can use an old computer system as a server." == Vulnerabilities == In September 2020, a hotfix was released to patch a security vulnerability that could leak sensitive information from the server environment since versions released in April 2019 (develop branch) and June 2019 (stable).
Wetware (brain)
Wetware is a term drawn from the computer-related idea of hardware or software, but applied to biological life forms. == Usage == The prefix "wet" is a reference to the water found in living creatures. Wetware is used to describe the elements equivalent to hardware and software found in a person, especially the central nervous system (CNS) and the human mind. The term wetware finds use in works of fiction, in scholarly publications and in popularizations. The "hardware" component of wetware concerns the bioelectric and biochemical properties of the CNS, specifically the brain. If the sequence of impulses traveling across the various neurons are thought of symbolically as software, then the physical neurons would be the hardware. The amalgamated interaction of this software and hardware is manifested through continuously changing physical connections, and chemical and electrical influences that spread across the body. The process by which the mind and brain interact to produce the collection of experiences that we define as self-awareness is in question. == History == Although the exact definition has shifted over time, the term Wetware and its fundamental reference to "the physical mind" has been around at least since the mid-1950s. Mostly used in relatively obscure articles and papers, it was not until the heyday of cyberpunk, however, that the term found broad adoption. Among the first uses of the term in popular culture was the Bruce Sterling novel Schismatrix (1985) and the Michael Swanwick novel Vacuum Flowers (1987). Rudy Rucker references the term in a number of books, including one entitled Wetware (1988): ... all sparks and tastes and tangles, all its stimulus/response patterns – the whole bio-cybernetic software of mind. Rucker did not use the word to simply mean a brain, nor in the human-resources sense of employees. He used wetware to stand for the data found in any biological system, analogous perhaps to the firmware that is found in a ROM chip. In Rucker's sense, a seed, a plant graft, an embryo, or a biological virus are all wetware. DNA, the immune system, and the evolved neural architecture of the brain are further examples of wetware in this sense. Rucker describes his conception in a 1992 compendium The Mondo 2000 User's Guide to the New Edge, which he quotes in a 2007 blog entry. Early cyber-guru Arthur Kroker used the term in his blog. With the term getting traction in trendsetting publications, it became a buzzword in the early 1990s. In 1991, Dutch media theorist Geert Lovink organized the Wetware Convention in Amsterdam, which was supposed to be an antidote to the "out-of-body" experiments conducted in high-tech laboratories, such as experiments in virtual reality. Timothy Leary, in an appendix to Info-Psychology originally written in 1975–76 and published in 1989, used the term wetware, writing that "psychedelic neuro-transmitters were the hot new technology for booting-up the 'wetware' of the brain". Another common reference is: "Wetware has 7 plus or minus 2 temporary registers." The numerical allusion is to a classic 1957 article by George A. Miller, The magical number 7 plus or minus two: some limits in our capacity for processing information, which later gave way to Miller's law.
Spike-and-slab regression
Spike-and-slab regression is a type of Bayesian linear regression in which a particular hierarchical prior distribution for the regression coefficients is chosen such that only a subset of the possible regressors is retained. The technique is particularly useful when the number of possible predictors is larger than the number of observations. The idea of the spike-and-slab model was originally proposed by Mitchell & Beauchamp (1988). The approach was further significantly developed by Madigan & Raftery (1994) and George & McCulloch (1997). A recent and important contribution to this literature is Ishwaran & Rao (2005). == Model description == Suppose we have P possible predictors in some model. Vector γ has a length equal to P and consists of zeros and ones. This vector indicates whether a particular variable is included in the regression or not. If no specific prior information on initial inclusion probabilities of particular variables is available, a Bernoulli prior distribution is a common default choice. Conditional on a predictor being in the regression, we identify a prior distribution for the model coefficient, which corresponds to that variable (β). A common choice on that step is to use a normal prior with a mean equal to zero and a large variance calculated based on ( X T X ) − 1 {\displaystyle (X^{T}X)^{-1}} (where X {\displaystyle X} is a design matrix of explanatory variables of the model). A draw of γ from its prior distribution is a list of the variables included in the regression. Conditional on this set of selected variables, we take a draw from the prior distribution of the regression coefficients (if γi = 1 then βi ≠ 0 and if γi = 0 then βi = 0). βγ denotes the subset of β for which γi = 1. In the next step, we calculate a posterior probability for both inclusion and coefficients by applying a standard statistical procedure. All steps of the described algorithm are repeated thousands of times using the Markov chain Monte Carlo (MCMC) technique. As a result, we obtain a posterior distribution of γ (variable inclusion in the model), β (regression coefficient values) and the corresponding prediction of y. The model got its name (spike-and-slab) due to the shape of the two prior distributions. The "spike" is the probability of a particular coefficient in the model to be zero. The "slab" is the prior distribution for the regression coefficient values. An advantage of Bayesian variable selection techniques is that they are able to make use of prior knowledge about the model. In the absence of such knowledge, some reasonable default values can be used; to quote Scott and Varian (2013): "For the analyst who prefers simplicity at the cost of some reasonable assumptions, useful prior information can be reduced to an expected model size, an expected R2, and a sample size ν determining the weight given to the guess at R2." Some researchers suggest the following default values: R2 = 0.5, ν = 0.01, and π = 0.5 (parameter of a prior Bernoulli distribution).
Algorithm selection
Algorithm selection (sometimes also called per-instance algorithm selection or offline algorithm selection) is a meta-algorithmic technique to choose an algorithm from a portfolio on an instance-by-instance basis. It is motivated by the observation that on many practical problems, different algorithms have different performance characteristics. That is, while one algorithm performs well in some scenarios, it performs poorly in others and vice versa for another algorithm. If we can identify when to use which algorithm, we can optimize for each scenario and improve overall performance. This is what algorithm selection aims to do. The only prerequisite for applying algorithm selection techniques is that there exists (or that there can be constructed) a set of complementary algorithms. == Definition == Given a portfolio P {\displaystyle {\mathcal {P}}} of algorithms A ∈ P {\displaystyle {\mathcal {A}}\in {\mathcal {P}}} , a set of instances i ∈ I {\displaystyle i\in {\mathcal {I}}} and a cost metric m : P × I → R {\displaystyle m:{\mathcal {P}}\times {\mathcal {I}}\to \mathbb {R} } , the algorithm selection problem consists of finding a mapping s : I → P {\displaystyle s:{\mathcal {I}}\to {\mathcal {P}}} from instances I {\displaystyle {\mathcal {I}}} to algorithms P {\displaystyle {\mathcal {P}}} such that the cost ∑ i ∈ I m ( s ( i ) , i ) {\displaystyle \sum _{i\in {\mathcal {I}}}m(s(i),i)} across all instances is optimized. == Examples == === Boolean satisfiability problem (and other hard combinatorial problems) === A well-known application of algorithm selection is the Boolean satisfiability problem. Here, the portfolio of algorithms is a set of (complementary) SAT solvers, the instances are Boolean formulas, the cost metric is for example average runtime or number of unsolved instances. So, the goal is to select a well-performing SAT solver for each individual instance. In the same way, algorithm selection can be applied to many other N P {\displaystyle {\mathcal {NP}}} -hard problems (such as mixed integer programming, CSP, AI planning, TSP, MAXSAT, QBF and answer set programming). Competition-winning systems in SAT are SATzilla, 3S and CSHC === Machine learning === In machine learning, algorithm selection is better known as meta-learning. The portfolio of algorithms consists of machine learning algorithms (e.g., Random Forest, SVM, DNN), the instances are data sets and the cost metric is for example the error rate. So, the goal is to predict which machine learning algorithm will have a small error on each data set. == Instance features == The algorithm selection problem is mainly solved with machine learning techniques. By representing the problem instances by numerical features f {\displaystyle f} , algorithm selection can be seen as a multi-class classification problem by learning a mapping f i ↦ A {\displaystyle f_{i}\mapsto {\mathcal {A}}} for a given instance i {\displaystyle i} . Instance features are numerical representations of instances. For example, we can count the number of variables, clauses, average clause length for Boolean formulas, or number of samples, features, class balance for ML data sets to get an impression about their characteristics. === Static vs. probing features === We distinguish between two kinds of features: Static features are in most cases some counts and statistics (e.g., clauses-to-variables ratio in SAT). These features ranges from very cheap features (e.g. number of variables) to very complex features (e.g., statistics about variable-clause graphs). Probing features (sometimes also called landmarking features) are computed by running some analysis of algorithm behavior on an instance (e.g., accuracy of a cheap decision tree algorithm on an ML data set, or running for a short time a stochastic local search solver on a Boolean formula). These feature often cost more than simple static features. === Feature costs === Depending on the used performance metric m {\displaystyle m} , feature computation can be associated with costs. For example, if we use running time as performance metric, we include the time to compute our instance features into the performance of an algorithm selection system. SAT solving is a concrete example, where such feature costs cannot be neglected, since instance features for CNF formulas can be either very cheap (e.g., to get the number of variables can be done in constant time for CNFs in the DIMACs format) or very expensive (e.g., graph features which can cost tens or hundreds of seconds). It is important to take the overhead of feature computation into account in practice in such scenarios; otherwise a misleading impression of the performance of the algorithm selection approach is created. For example, if the decision which algorithm to choose can be made with perfect accuracy, but the features are the running time of the portfolio algorithms, there is no benefit to the portfolio approach. This would not be obvious if feature costs were omitted. == Approaches == === Regression approach === One of the first successful algorithm selection approaches predicted the performance of each algorithm m ^ A : I → R {\displaystyle {\hat {m}}_{\mathcal {A}}:{\mathcal {I}}\to \mathbb {R} } and selected the algorithm with the best predicted performance a r g min A ∈ P m ^ A ( i ) {\displaystyle arg\min _{{\mathcal {A}}\in {\mathcal {P}}}{\hat {m}}_{\mathcal {A}}(i)} for an instance i {\displaystyle i} . === Clustering approach === A common assumption is that the given set of instances I {\displaystyle {\mathcal {I}}} can be clustered into homogeneous subsets and for each of these subsets, there is one well-performing algorithm for all instances in there. So, the training consists of identifying the homogeneous clusters via an unsupervised clustering approach and associating an algorithm with each cluster. A new instance is assigned to a cluster and the associated algorithm selected. A more modern approach is cost-sensitive hierarchical clustering using supervised learning to identify the homogeneous instance subsets. === Pairwise cost-sensitive classification approach === A common approach for multi-class classification is to learn pairwise models between every pair of classes (here algorithms) and choose the class that was predicted most often by the pairwise models. We can weight the instances of the pairwise prediction problem by the performance difference between the two algorithms. This is motivated by the fact that we care most about getting predictions with large differences correct, but the penalty for an incorrect prediction is small if there is almost no performance difference. Therefore, each instance i {\displaystyle i} for training a classification model A 1 {\displaystyle {\mathcal {A}}_{1}} vs A 2 {\displaystyle {\mathcal {A}}_{2}} is associated with a cost | m ( A 1 , i ) − m ( A 2 , i ) | {\displaystyle |m({\mathcal {A}}_{1},i)-m({\mathcal {A}}_{2},i)|} . == Requirements == The algorithm selection problem can be effectively applied under the following assumptions: The portfolio P {\displaystyle {\mathcal {P}}} of algorithms is complementary with respect to the instance set I {\displaystyle {\mathcal {I}}} , i.e., there is no single algorithm A ∈ P {\displaystyle {\mathcal {A}}\in {\mathcal {P}}} that dominates the performance of all other algorithms over I {\displaystyle {\mathcal {I}}} (see figures to the right for examples on complementary analysis). In some application, the computation of instance features is associated with a cost. For example, if the cost metric is running time, we have also to consider the time to compute the instance features. In such cases, the cost to compute features should not be larger than the performance gain through algorithm selection. == Application domains == Algorithm selection is not limited to single domains but can be applied to any kind of algorithm if the above requirements are satisfied. Application domains include: hard combinatorial problems: SAT, Mixed Integer Programming, CSP, AI Planning, TSP, MAXSAT, QBF and Answer Set Programming combinatorial auctions in machine learning, the problem is known as meta-learning software design black-box optimization multi-agent systems numerical optimization linear algebra, differential equations evolutionary algorithms vehicle routing problem power systems For an extensive list of literature about algorithm selection, we refer to a literature overview. == Variants of algorithm selection == === Online selection === Online algorithm selection refers to switching between different algorithms during the solving process. This is useful as a hyper-heuristic. In contrast, offline algorithm selection selects an algorithm for a given instance only once and before the solving process. === Computation of schedules === An extension of algorithm selection is the per-instance algorithm scheduling problem, in which we do not select only one solver, but we select a time budget for each algorithm
IPUMS
IPUMS, originally the Integrated Public Use Microdata Series, is the world's largest individual-level population database. IPUMS consists of microdata samples from United States (IPUMS-USA) and international (IPUMS-International) census records, as well as data from U.S. and international surveys. The records are converted into a consistent format and made available to researchers through a web-based data dissemination and analysis system. IPUMS is housed at the Institute for Social Research and Data Innovation (ISRDI), an interdisciplinary research center at the University of Minnesota, under the direction of Professor Steven Ruggles. == Description == IPUMS includes all persons enumerated in the United States censuses from 1850 to 1950 (though, the 1890 census is missing because it was destroyed in a fire) and from the American Community Survey since 2000 and the Current Population Survey since 1962. IPUMS includes household-level data for United States Censuses from 1790 to 1840, due to the first six censuses only including the name of the head of household, with tallied household totals following. IPUMS provides consistent variable names, coding schemes, and documentation across all the samples, facilitating the analysis of long-term change. IPUMS-International includes countries from Africa, Asia, Europe, and Latin America for 1960 forward. The database currently includes more than a billion individuals enumerated in 365 censuses from 94 countries around the world. IPUMS-International converts census microdata for multiple countries into a consistent format, allowing for comparisons across countries and time periods. Special efforts are made to simplify use of the data while losing no meaningful information. Comprehensive documentation is provided in a coherent form to facilitate comparative analyses of social and economic change. Additional databases in the IPUMS family include the: North Atlantic Population Project (NAPP) IPUMS National Historical Geographic Information System (NHGIS) IPUMS Health Surveys IPUMS Global Health IPUMS Time Use The Journal of American History described the effort as "One of the great archival projects of the past two decades." Liens Socio, the French portal for the social sciences, gave IPUMS the only “best site” designation that has gone to any non-French website, writing “IPUMS est un projet absolument extraordinaire...époustouflante [mind-blowing]!” The official motto of IPUMS is "use it for good, never for evil." All public IPUMS data and documentation are available online free of charge.
Cost-sensitive machine learning
Cost-sensitive machine learning is an approach within machine learning that considers varying costs associated with different types of errors. This method diverges from traditional approaches by introducing a cost matrix, explicitly specifying the penalties or benefits for each type of prediction error. The inherent difficulty which cost-sensitive machine learning tackles is that minimizing different kinds of classification errors is a multi-objective optimization problem. == Overview == Cost-sensitive machine learning optimizes models based on the specific consequences of misclassifications, making it a valuable tool in various applications. It is especially useful in problems with a high imbalance in class distribution and a high imbalance in associated costs Cost-sensitive machine learning introduces a scalar cost function in order to find one (of multiple) Pareto optimal points in this multi-objective optimization problem (similar to the Weighted sum model) == Cost Matrix == The cost matrix is a crucial element within cost-sensitive modeling, explicitly defining the costs or benefits associated with different prediction errors in classification tasks. Represented as a table, the matrix aligns true and predicted classes, assigning a cost value to each combination. For instance, in binary classification, it may distinguish costs for false positives and false negatives. The utility of the cost matrix lies in its application to calculate the expected cost or loss. The formula, expressed as a double summation, utilizes joint probabilities: Expected Loss = ∑ i ∑ j P ( Actual i , Predicted j ) ⋅ Cost Actual i , Predicted j {\displaystyle {\text{Expected Loss}}=\sum _{i}\sum _{j}P({\text{Actual}}_{i},{\text{Predicted}}_{j})\cdot {\text{Cost}}_{{\text{Actual}}_{i},{\text{Predicted}}_{j}}} Here, P ( Actual i , Predicted j ) {\displaystyle P({\text{Actual}}_{i},{\text{Predicted}}_{j})} denotes the joint probability of actual class i {\displaystyle i} and predicted class j {\displaystyle j} , providing a nuanced measure that considers both the probabilities and associated costs. This approach allows practitioners to fine-tune models based on the specific consequences of misclassifications, adapting to scenarios where the impact of prediction errors varies across classes. == Applications == === Fraud Detection === In the realm of data science, particularly in finance, cost-sensitive machine learning is applied to fraud detection. By assigning different costs to false positives and false negatives, models can be fine-tuned to minimize the overall financial impact of misclassifications. === Medical Diagnostics === In healthcare, cost-sensitive machine learning plays a role in medical diagnostics. The approach allows for customization of models based on the potential harm associated with misdiagnoses, ensuring a more patient-centric application of machine learning algorithms. == Challenges == A typical challenge in cost-sensitive machine learning is the reliable determination of the cost matrix which may evolve over time. == Literature == Cost-Sensitive Machine Learning. USA, CRC Press, 2011. ISBN 9781439839287 Abhishek, K., Abdelaziz, D. M. (2023). Machine Learning for Imbalanced Data: Tackle Imbalanced Datasets Using Machine Learning and Deep Learning Techniques. (n.p.): Packt Publishing. ISBN 9781801070881