Grokipedia is an AI-generated online encyclopedia operated by the American company xAI. The site was launched on October 27, 2025. Some entries are generated by Grok, a large language model owned by the same company, while others were forked from Wikipedia, with some altered and some used nearly verbatim. Articles cannot be directly edited, though logged-in visitors to the encyclopedia can suggest new articles or corrections via a pop-up form, which are reviewed by Grok. The xAI founder Elon Musk suggested Grokipedia could be an alternative to Wikipedia that would "purge out the propaganda" he believes is promoted by the latter, describing Wikipedia as "woke" and an "extension of legacy media propaganda". External analysis of Grokipedia's content has focused on its accuracy and biases due to hallucinations and potential algorithmic bias, which reviewers have described as promoting right-wing perspectives and Musk's views. The majority of coverage has described the website as validating, promoting, and legitimizing a variety of debunked conspiracy theories and ideas against scientific consensus on topics such as HIV/AIDS denialism, vaccines and autism, climate change, and race and intelligence. The site has been accused of whitewashing far-right extremism, such as by falsely claiming a white genocide is actively occurring. Several right-wing figures have welcomed the site. Studies have highlighted its use of sources deemed as having very low credibility such as X conversations and neo-Nazi websites, and for writing about far-right figures and topics in a promotional manner. == Background == Wikipedia is an online encyclopedia written and maintained by a community of volunteers. Its possible bias has been studied and debated. In 2018, Haaretz noted "Wikipedia has succeeded in being accused of being both too liberal and too conservative, and has critics from across the spectrum". xAI is an American AI company founded by Elon Musk in 2023. Its flagship product is the family of large language models called Grok. == History == In 2021, Musk expressed affection for Wikipedia on its 20th anniversary. In 2022, however, Musk argued that Wikipedia was "losing its objectivity", and in 2023, said he would donate US$1 billion to the project if it was pejoratively renamed "Dickipedia". In December 2024, Musk called for a boycott of donations to Wikipedia over its perceived left-wing bias, calling it "Wokepedia". In January 2025, Musk made a series of statements on Twitter denouncing Wikipedia for its description of the incident where he made a controversial gesture, which many viewed as resembling a Nazi salute, at president Donald Trump's second inauguration. Musk has since positioned Grokipedia as an alternative to Wikipedia that would "purge out the propaganda" in the latter, with Musk describing Wikipedia as "woke" and an "extension of legacy media propaganda". === Idea and announcement === In September 2025, Musk spoke at the All-In podcast conference with David O. Sacks, the White House advisor on AI and cryptocurrency, about how Grok consumed data from Wikipedia and other sources to gain more complete knowledge of the world. Sacks suggested publishing its knowledge base as an artifact called "Grokipedia", saying "Wikipedia is so biased, it's a constant war". Following the conversation, Musk announced that xAI was building a new AI-generated online encyclopedia called Grokipedia. According to Musk's announcement, it would be an AI-powered knowledge base designed to rival Wikipedia by addressing its perceived biases, errors, and ideological slants. The project positioned itself within a history of ideologically driven alternatives to Wikipedia, such as the conservative Conservapedia (launched in 2006) and the Russian-government-friendly Ruwiki (launched in 2023). However, Grokipedia is distinct in its core reliance on artificial intelligence rather than human community editing. === Launch and traffic === On October 6, 2025, Musk announced that the early version of Grokipedia was scheduled for release in two weeks, but the project was postponed briefly to address content quality issues. It launched on October 27, 2025, labeled "v 0.1", with over 800,000 articles, compared to over seven million English Wikipedia articles as of September 1, 2025. According to an initial analysis of usage figures by Similarweb, which evaluates data from registered users and partners, Grokipedia recorded a peak of over 460,000 website visits in the US on October 28, 2025. After that, traffic dropped significantly and settled at around 35,000 visits per day between November 8 and 11, 2025. As of early 2026, it had over 5.6 million articles. In January 2026, The Guardian reported that GPT-5.2 frequently cited Grokipedia as a source in responses, raising concerns of misinformation on ChatGPT. The same month, The Verge reported that Google's AI Overviews, AI Mode, and Gemini language model, as well as Microsoft Copilot and Perplexity AI, used Grokipedia to answer niche, obscure, or highly specific factual questions or "non-sensitive queries." According to a case study published by SEO Engico, the site received only 19 clicks from Google Search in November 2025 but reached approximately 3.2 million monthly clicks by January 2026, with over 900,000 pages indexed and millions of ranking keywords. Analysts attributed the surge in part to the site's technical structure and large-scale AI-generated content production. In early February 2026, Grokipedia's visibility in Google Search declined sharply. SEO analysts, including Glenn Gabe and Malte Landwehr, reported a significant drop in rankings across Google organic results as well as in Google AI Overviews and AI Mode. The same case study cited independent reviews that identified citation quality concerns, including references to low-credibility sources and instances of self-citation. By mid-February 2026, Grokipedia had reportedly lost much of its previous search visibility, and Wikipedia ranked above it for searches related to its own name. === Updates === ==== Future ==== In November 2025, Musk announced that he eventually plans to change the name of the site to Encyclopedia Galactica when Grokipedia is "good enough", saying that it had a "long way to go". This name is taken from the publication of that title in the works of Isaac Asimov and Douglas Adams. Musk said that he hoped to send copies of the encyclopedia to "the Moon and Mars and out to deep space". == Content == The Grok large language model generates and fact-checks articles on Grokipedia. Users cannot directly edit Grokipedia articles, but logged-in users can suggest edits and report errors, with such submissions being reviewed and implemented by the Grok AI. Some articles are nearly identical to their Wikipedia entries, but the format of Grokipedia citations is different, and some Grokipedia articles were republished almost verbatim, accompanied by a disclaimer noting that the content was "adapted from Wikipedia" under a Creative Commons license. Others were completely rewritten from scratch using Musk's AI chatbot, Grok. Forbes identified the articles AMD, Lamborghini, and PlayStation 5 as examples of copied Wikipedia articles. Articles attributed to Wikipedia carry a Creative Commons Attribution-ShareAlike license, while the license of other articles is licensed under the "X Community License", a license that accepts reuse and remixing for "non-commercial and research purposes" and commercial use that abides to "all of the guardrails provided in xAI's Acceptable Use Policy". On October 31, 2025, Musk clarified that the duplication of Wikipedia articles was intentional, saying that the Grokipedia team instructed Grok to compile Wikipedia's top 1 million articles and make content changes to them. The site's design has been described as minimalist with a simple homepage including little more than a large search bar. In a comparative textual analysis of the most heavily edited matched article pairs from Grokipedia and Wikipedia, Grokipedia entries are substantially longer and less densely referenced, indicating that AI-produced encyclopedias prioritize exposition rather than source-based validation. Starting in version 0.2, Grok reviews and implements approved suggested edits, and a small panel rotates through a display of the names of several recently edited articles. In February 2026, the Columbia Journalism Review reported on an analysis by the Tow Center for Digital Journalism finding that Grok, the AI behind Grokipedia, had increasingly begun suggesting and approving edits to the site itself without human involvement. According to the report, AI-generated edit suggestions overtook human submissions in December 2025 and accounted for more than three-quarters of proposed changes. The analysis raised concerns about transparency, editorial oversight, and fact-checking standards, particularly after instances in which Grok proposed or modified politically s
Personal cloud
A personal cloud is a collection of digital content and services that are accessible from any device through the Internet. It is not a tangible entity, but a place that gives users the ability to store, synchronize, stream and share content on a relative core, moving from one platform, screen and location to another. Created on connected services and applications, it reflects and sets consumer expectations for how next-generation computing services will work. The four primary types of personal cloud in use today are: Online cloud, NAS device cloud, server device cloud, and home-made clouds. == Online cloud == The online cloud is sometimes referred to as the public cloud. It is the cloud computing model where online resources like software and data storage are made available over the Internet. Typically, an individual or organization has little control over the ecosystem in which the online cloud is hosted, and the core infrastructure is shared between many individuals and organizations. The data and applications provided by the service provider are logically segregated so that only those authorized are allowed access. == NAS device cloud == A network-attached storage (NAS) device is a computer connected to a network that provides only file-based data storage services to other devices on the network. Although it may technically be possible to run other software on a NAS device, it is not designed to be a general purpose server. Cloud NAS is remote storage that is accessed over the Internet as if it were local. A cloud NAS is often used for backups and archiving. One of the benefits of NAS Cloud is that data in the cloud can be accessed at any time from anywhere. The main drawback, however, is that the speed of the transfer rate is only as fast as the network connection the data is accessed over and can therefore be fairly slow. == Server device cloud == In many ways cloud servers work in the same way as physical servers but the functions they perform can be very different. Typically, the cloud server is an on-premises device that is connected to the Internet and gives users the functions available on the online cloud but with the added benefit and security of the files being in their control on their premises. The server cloud has been historically enterprise-based deployed by businesses needing an in-house cloud. However, there are also in-house options available for individual users. == Home-made clouds == For the more technologically proficient user a common solution for using a personal cloud is to create a home-made cloud system by connecting an external USB hard drive to a Wi-Fi router. This enables both wired and wireless computers to access the USB hard drive and use it for storage or for retrieving files a user needs to share on the network thereby acting like a cloud. Setting up a personal cloud requires a user to have particular skills in technology and network setup. One of the risks associated with improper setup is security, and leaving the files accessible to anyone with technical knowledge. Not every router supports this type of access and modification.
Targeted maximum likelihood estimation
Targeted Maximum Likelihood Estimation (TMLE) (also more accurately referred to as Targeted Minimum Loss-Based Estimation) is a general statistical estimation framework for causal inference and semiparametric models. TMLE combines ideas from maximum likelihood estimation, semiparametric efficiency theory, and machine learning. It was introduced by Mark J. van der Laan and colleagues in the mid-2000s as a method that yields asymptotically efficient plug-in estimators while allowing the use of flexible, data-adaptive algorithms such as ensemble machine learning for nuisance parameter estimation. TMLE is used in epidemiology, biostatistics, and the social sciences to estimate causal effects in observational and experimental studies. Applications of TMLE include Longitudinal TMLE (LTMLE) for time-varying treatments and confounders. Variations in how the targeting step in TMLE is carried out have resulted in various versions of TMLE such as Collaborative TMLE (CTMLE) and Adaptive TMLE for improved finite-sample performance and automated variable selection. == History == The TMLE framework was first described by van der Laan and Rubin (2006) as a general approach for the construction of efficient plug-in estimators of smooth features of the data density. It was demonstrated in the context of causal inference and missing data problems. It was developed to address limitations of traditional doubly robust methods, such as Augmented Inverse Probability Weighting (AIPW), by respecting the plug-in principle in the sense that it respects that the target parameter is a function of the data density that is an element of the statistical model. TMLE estimates the data density or relevant parts of it with machine learning and targets these machine learning fits before it is plugged in the target parameter mapping. In this manner, a TMLE always respects global knowledge and satisfies known bounds such as that the target parameter is a probability . Since its introduction, TMLE has been developed in a series of theoretical and applied papers, culminating in book-length treatments of the method and its applications to survival analysis, adaptive designs, and longitudinal data. == Methodology == At its core, TMLE is a two-step estimation procedure: Initial estimation: Machine learning methods (such as the Super Learner ensemble) are used to obtain flexible estimates of nuisance parameters, such as outcome regressions and propensity scores. Targeting step: The initial estimate is updated by solving a score equation (the efficient influence function) so that the final estimator is consistent, asymptotically normal, and efficient under mild regularity conditions. The targeted machine learning fit is then mapped into the corresponding estimator of the target parameter by simply plugging it in the target parameter mapping. This approach balances the bias–variance trade-off by combining data-adaptive estimation with semiparametric efficiency theory. TMLE is doubly robust, meaning it remains consistent if either the outcome model or the treatment model is consistently estimated. === Formula === Here we explain the TMLE of the average treatment effect of a binary treatment on an outcome adjusting for baseline covariates. Consider i.i.d. observations O i = ( W i , A i , Y i ) {\displaystyle O_{i}=(W_{i},A_{i},Y_{i})} from a distribution P 0 {\displaystyle P_{0}} , where W {\displaystyle W} are baseline covariates, A {\displaystyle A} is a binary treatment, and Y {\displaystyle Y} is an outcome. Let Q ¯ ( a , w ) = E [ Y ∣ A = a , W = w ] {\displaystyle {\bar {Q}}(a,w)=\mathbb {E} [Y\mid A=a,W=w]} represent the outcome model and g ( a ∣ w ) = P ( A = a ∣ W = w ) {\displaystyle g(a\mid w)=P(A=a\mid W=w)} represent the propensity score. The average treatment effect (ATE) is given by ψ 0 = E { Q ¯ ( 1 , W ) − Q ¯ ( 0 , W ) } . {\displaystyle \psi _{0}=\mathbb {E} \{{\bar {Q}}(1,W)-{\bar {Q}}(0,W)\}.} A basic TMLE for the ATE proceeds as follows: Step 1: Estimate initial models. Obtain estimates Q ¯ ^ ( a , w ) {\displaystyle {\hat {\bar {Q}}}(a,w)} and g ^ ( a ∣ w ) {\displaystyle {\hat {g}}(a\mid w)} , often using flexible methods such as Super Learner. Step 2: Compute the clever covariate. Define: H ( A , W ) = A g ^ ( 1 ∣ W ) − 1 − A g ^ ( 0 ∣ W ) . {\displaystyle H(A,W)={\frac {A}{{\hat {g}}(1\mid W)}}-{\frac {1-A}{{\hat {g}}(0\mid W)}}.} Step 3: Estimate the fluctuation parameter. Fit a logistic regression of Y {\displaystyle Y} on H ( A , W ) {\displaystyle H(A,W)} with logit ( Q ¯ ^ ( A , W ) ) {\displaystyle \operatorname {logit} ({\hat {\bar {Q}}}(A,W))} as offset. This yields ε ^ {\displaystyle {\hat {\varepsilon }}} , the MLE that solves the score equation: 1 n ∑ i = 1 n H ( A i , W i ) { Y i − Q ¯ ^ ε ( A i , W i ) } = 0. {\displaystyle {\frac {1}{n}}\sum _{i=1}^{n}H(A_{i},W_{i}){\big \{}Y_{i}-{\hat {\bar {Q}}}^{\varepsilon }(A_{i},W_{i}){\big \}}=0.} Step 4: Update the initial estimate. Apply the "blip" to obtain the targeted estimate: Q ¯ ^ ∗ ( A , W ) = expit ( logit ( Q ¯ ^ ( A , W ) ) + ε ^ H ( A , W ) ) . {\displaystyle {\hat {\bar {Q}}}^{}(A,W)=\operatorname {expit} {\Big (}\operatorname {logit} {\big (}{\hat {\bar {Q}}}(A,W){\big )}+{\hat {\varepsilon }}\,H(A,W){\Big )}.} Step 5: Compute the TMLE. The ATE estimate is: ψ ^ TMLE = 1 n ∑ i = 1 n [ Q ¯ ^ ∗ ( 1 , W i ) − Q ¯ ^ ∗ ( 0 , W i ) ] . {\displaystyle {\hat {\psi }}_{\text{TMLE}}={\frac {1}{n}}\sum _{i=1}^{n}{\big [}{\hat {\bar {Q}}}^{}(1,W_{i})-{\hat {\bar {Q}}}^{}(0,W_{i}){\big ]}.} Inference. The efficient influence function (EIF) for the ATE is: D ∗ ( O ) = H ( A , W ) { Y − Q ¯ ∗ ( A , W ) } + Q ¯ ∗ ( 1 , W ) − Q ¯ ∗ ( 0 , W ) − ψ . {\displaystyle D^{}(O)=H(A,W)\{Y-{\bar {Q}}^{}(A,W)\}+{\bar {Q}}^{}(1,W)-{\bar {Q}}^{}(0,W)-\psi .} The variance is estimated by σ ^ 2 = n − 1 ∑ i = 1 n ( D ∗ ( O i ) ) 2 {\displaystyle {\hat {\sigma }}^{2}=n^{-1}\sum _{i=1}^{n}{\big (}D^{}(O_{i}){\big )}^{2}} , yielding Wald-type confidence intervals ψ ^ TMLE ± z 1 − α / 2 σ ^ / n {\displaystyle {\hat {\psi }}_{\text{TMLE}}\pm z_{1-\alpha /2}\,{\hat {\sigma }}/{\sqrt {n}}} . Remark. For continuous outcomes, a linear fluctuation Q ¯ ^ ∗ = Q ¯ ^ + ε ^ H {\displaystyle {\hat {\bar {Q}}}^{}={\hat {\bar {Q}}}+{\hat {\varepsilon }}\,H} may be used instead. For bounded continuous outcomes, the logistic fluctuation (after rescaling Y {\displaystyle Y} to [ 0 , 1 ] {\displaystyle [0,1]} ) is often preferred for improved finite-sample performance. == Applications == TMLE has been applied in: Epidemiology: Estimating causal effects of exposures and interventions in observational cohort studies. Clinical trials and real-world evidence: The Targeted Learning roadmap provides a structured framework for generating and validating real-world evidence (RWE), bridging randomized trials and observational data using TMLE and related estimation techniques. This approach enables transparency, sensitivity analysis, and stronger causal inference for regulatory and clinical trial contexts. High-dimensional settings: Integration with ensemble methods for causal effect estimation. TMLE has been successfully applied in pharmacoepidemiology where a large number of covariates are automatically selected to adjust for confounding. In a study of post–myocardial infarction statin use and 1-year mortality, TMLE demonstrated robust performance relative to inverse probability weighting in scenarios with hundreds of potential confounders. == Derivatives and extensions == Longitudinal TMLE (LTMLE): A methodological extension of TMLE for longitudinal data with time-varying treatments, confounders, and censoring. It allows the estimation of dynamic treatment regimes and intervention-specific causal effects over time. This framework was originally introduced by van der Laan & Gruber (2012). Collaborative TMLE (CTMLE): Enhances finite-sample performance and variable selection by collaboratively fitting the treatment mechanism in conjunction with the target parameter. == Software == Several R packages implement TMLE and related methods: tmle: Functions for binary, categorical, and continuous outcomes. ltmle: Implementation for longitudinal data with time-varying treatments and outcomes. ctmle: Algorithms for collaborative TMLE and adaptive variable selection. SuperLearner: A theoretically grounded, cross-validated ensemble learning method that combines predictions from multiple algorithms to minimize predictive risk. Widely used in TMLE for estimating nuisance parameters. The original implementation is available as the R package SuperLearner. Recent machine learning platforms like H2O AutoML implement similar ensemble strategies, combining diverse learners in parallel and leveraging stacking and blending techniques, effectively functioning as a large-scale Super Learner.
Cross-entropy
In information theory, the cross-entropy between two probability distributions p {\displaystyle p} and q {\displaystyle q} , over the same underlying set of events, measures the average number of bits needed to identify an event drawn from the set when the coding scheme used for the set is optimized for an estimated probability distribution q {\displaystyle q} , rather than the true distribution p {\displaystyle p} . == Definition == The cross-entropy of the distribution q {\displaystyle q} relative to a distribution p {\displaystyle p} over a given set is defined as follows: H ( p , q ) = − E p [ log q ] , {\displaystyle H(p,q)=-\operatorname {E} _{p}[\log q],} where E p [ ⋅ ] {\displaystyle \operatorname {E} _{p}[\cdot ]} is the expected value operator with respect to the distribution p {\displaystyle p} . The definition may be formulated using the Kullback–Leibler divergence D K L ( p ∥ q ) {\displaystyle D_{\mathrm {KL} }(p\parallel q)} , divergence of p {\displaystyle p} from q {\displaystyle q} (also known as the relative entropy of p {\displaystyle p} with respect to q {\displaystyle q} ). H ( p , q ) = H ( p ) + D K L ( p ∥ q ) , {\displaystyle H(p,q)=H(p)+D_{\mathrm {KL} }(p\parallel q),} where H ( p ) {\displaystyle H(p)} is the entropy of p {\displaystyle p} . For discrete probability distributions p {\displaystyle p} and q {\displaystyle q} with the same support X {\displaystyle {\mathcal {X}}} , this means The situation for continuous distributions is analogous. We have to assume that p {\displaystyle p} and q {\displaystyle q} are absolutely continuous with respect to some reference measure r {\displaystyle r} (usually r {\displaystyle r} is a Lebesgue measure on a Borel σ-algebra). Let P {\displaystyle P} and Q {\displaystyle Q} be probability density functions of p {\displaystyle p} and q {\displaystyle q} with respect to r {\displaystyle r} . Then − ∫ X P ( x ) log Q ( x ) d x = E p [ − log Q ] , {\displaystyle -\int _{\mathcal {X}}P(x)\,\log Q(x)\,\mathrm {d} x=\operatorname {E} _{p}[-\log Q],} and therefore NB: The notation H ( p , q ) {\displaystyle H(p,q)} is also used for a different concept, the joint entropy of p {\displaystyle p} and q {\displaystyle q} . == Motivation == In information theory, the Kraft–McMillan theorem establishes that any directly decodable coding scheme for coding a message to identify one value x i {\displaystyle x_{i}} out of a set of possibilities { x 1 , … , x n } {\displaystyle \{x_{1},\ldots ,x_{n}\}} can be seen as representing an implicit probability distribution q ( x i ) = ( 1 2 ) ℓ i {\displaystyle q(x_{i})=\left({\frac {1}{2}}\right)^{\ell _{i}}} over { x 1 , … , x n } {\displaystyle \{x_{1},\ldots ,x_{n}\}} , where ℓ i {\displaystyle \ell _{i}} is the length of the code for x i {\displaystyle x_{i}} in bits. Therefore, cross-entropy can be interpreted as the expected message-length per datum when a wrong distribution q {\displaystyle q} is assumed while the data actually follows a distribution p {\displaystyle p} . That is why the expectation is taken over the true probability distribution p {\displaystyle p} and not q . {\displaystyle q.} Indeed the expected message-length under the true distribution p {\displaystyle p} is E p [ ℓ ] = − E p [ ln q ( x ) ln ( 2 ) ] = − E p [ log 2 q ( x ) ] = − ∑ x i p ( x i ) log 2 q ( x i ) = − ∑ x p ( x ) log 2 q ( x ) = H ( p , q ) . {\displaystyle {\begin{aligned}\operatorname {E} _{p}[\ell ]&=-\operatorname {E} _{p}\left[{\frac {\ln {q(x)}}{\ln(2)}}\right]\\[1ex]&=-\operatorname {E} _{p}\left[\log _{2}{q(x)}\right]\\[1ex]&=-\sum _{x_{i}}p(x_{i})\,\log _{2}q(x_{i})\\[1ex]&=-\sum _{x}p(x)\,\log _{2}q(x)=H(p,q).\end{aligned}}} == Estimation == There are many situations where cross-entropy needs to be measured but the distribution of p {\displaystyle p} is unknown. An example is language modeling, where a model is created based on a training set T {\displaystyle T} , and then its cross-entropy is measured on a test set to assess how accurate the model is in predicting the test data. In this example, p {\displaystyle p} is the true distribution of words in any corpus, and q {\displaystyle q} is the distribution of words as predicted by the model. Since the true distribution is unknown, cross-entropy cannot be directly calculated. In these cases, an estimate of cross-entropy is calculated using the following formula: H ( T , q ) = − ∑ i = 1 N 1 N log 2 q ( x i ) {\displaystyle H(T,q)=-\sum _{i=1}^{N}{\frac {1}{N}}\log _{2}q(x_{i})} where N {\displaystyle N} is the size of the test set, and q ( x ) {\displaystyle q(x)} is the probability of event x {\displaystyle x} estimated from the training set. In other words, q ( x i ) {\displaystyle q(x_{i})} is the probability estimate of the model that the i-th word of the text is x i {\displaystyle x_{i}} . The sum is averaged over the N {\displaystyle N} words of the test. This is a Monte Carlo estimate of the true cross-entropy, where the test set is treated as samples from p ( x ) {\displaystyle p(x)} . == Relation to maximum likelihood == The cross entropy arises in classification problems when introducing a logarithm in the guise of the log-likelihood function. This section concerns the estimation of the probabilities of different discrete outcomes. To this end, denote a parametrized family of distributions by q θ {\displaystyle q_{\theta }} , with θ {\displaystyle \theta } subject to the optimization effort. Consider a given finite sequence of N {\displaystyle N} values x i {\displaystyle x_{i}} from a training set, obtained from conditionally independent sampling. The likelihood assigned to any considered parameter θ {\displaystyle \theta } of the model is then given by the product over all probabilities q θ ( X = x i ) {\displaystyle q_{\theta }(X=x_{i})} . Repeated occurrences are possible, leading to equal factors in the product. If the count of occurrences of the value equal to x {\displaystyle x} is denoted by # x {\displaystyle \#x} , then the frequency of that value equals # x / N {\displaystyle \#x/N} . If p ( X = x ) {\displaystyle p(X=x)} is the underlying probability distribution, for large N {\displaystyle N} we expect p ( X = x ) ≈ # x / N {\displaystyle p(X=x)\approx \#x/N} , by the law of large numbers. Writing our likelihood function as the product of observations from the distribution q θ {\displaystyle q_{\theta }} : L ( θ ; x ) = ∏ i q θ ( X = x i ) = ∏ x q θ ( X = x ) # x ≈ ∏ x q θ ( X = x ) N ⋅ p ( X = x ) = exp log [ ∏ x q θ ( X = x ) N ⋅ p ( X = x ) ] = exp ( ∑ x N ⋅ p ( X = x ) log q θ ( X = x ) ) , {\displaystyle {\begin{aligned}{\mathcal {L}}(\theta ;{\mathbf {x} })&=\prod _{i}q_{\theta }(X=x_{i})=\prod _{x}q_{\theta }(X=x)^{\#x}\\&\approx \prod _{x}q_{\theta }(X=x)^{N\cdot p(X=x)}=\exp \log \left[\prod _{x}q_{\theta }(X=x)^{N\cdot p(X=x)}\right]\\&=\exp \left(\sum _{x}N\cdot p(X=x)\log q_{\theta }(X=x)^{}\right),\end{aligned}}} where we have used the calculation rules for the logarithm in the final line. Notice how the exponent contains a − H ( p , q θ ) {\displaystyle -H(p,q_{\theta })} term. Taking the logarithm of both sides gives: log L ( θ ; x ) = − N ⋅ H ( p , q θ ) . {\displaystyle \log {\mathcal {L}}(\theta ;{\mathbf {x} })=-N\cdot H(p,q_{\theta }).} Since the logarithm is a monotonically increasing function, the maximizing value of θ {\displaystyle \theta } is unaffected by this final step. Similarly, the maximizing value of θ {\displaystyle \theta } is unaffected by the factor of N {\displaystyle N} . So we observe that the likelihood maximization amounts to minimization of the cross-entropy. == Cross-entropy minimization == Cross-entropy minimization is frequently used in optimization and rare-event probability estimation. When comparing a distribution q {\displaystyle q} against a fixed reference distribution p {\displaystyle p} , cross-entropy and KL divergence are identical up to an additive constant (since p {\displaystyle p} is fixed): According to the Gibbs' inequality, both take on their minimal values when p = q {\displaystyle p=q} , which is 0 {\displaystyle 0} for KL divergence, and H ( p ) {\displaystyle \mathrm {H} (p)} for cross-entropy. In the engineering literature, the principle of minimizing KL divergence (Kullback's "Principle of Minimum Discrimination Information") is often called the Principle of Minimum Cross-Entropy (MCE), or Minxent. However, as discussed in the article Kullback–Leibler divergence, sometimes the distribution q {\displaystyle q} is the fixed prior reference distribution, and the distribution p {\displaystyle p} is optimized to be as close to q {\displaystyle q} as possible, subject to some constraint. In this case the two minimizations are not equivalent. This has led to some ambiguity in the literature, with some authors attempting to resolve the inconsistency by restating cross-entropy to be D K L ( p ∥ q ) {\displaystyle D_{\mathrm {KL} }(p\parallel q)} , rather than H (
Teaching dimension
In computational learning theory, the teaching dimension of a concept class C is defined to be max c ∈ C { w C ( c ) } {\displaystyle \max _{c\in C}\{w_{C}(c)\}} , where w C ( c ) {\displaystyle {w_{C}(c)}} is the minimum size of a witness set for c in C. Intuitively, this measures the number of instances that are needed to identify a concept in the class, using supervised learning with examples provided by a helpful teacher who is trying to convey the concept as succinctly as possible. This definition was formulated in 1995 by Sally Goldman and Michael Kearns, based on earlier work by Goldman, Ron Rivest, and Robert Schapire. The teaching dimension of a finite concept class can be used to give a lower and an upper bound on the membership query cost of the concept class. In Stasys Jukna's book "Extremal Combinatorics", a lower bound is given for the teaching dimension in general: Let C be a concept class over a finite domain X. If the size of C is greater than 2 k ( | X | k ) , {\displaystyle 2^{k}{|X| \choose k},} then the teaching dimension of C is greater than k. However, there are more specific teaching models that make assumptions about teacher or learner, and can get lower values for the teaching dimension. For instance, several models are the classical teaching (CT) model, the optimal teacher (OT) model, recursive teaching (RT), preference-based teaching (PBT), and non-clashing teaching (NCT).
IWork
iWork is an office suite of applications created by Apple for its macOS, iPadOS, and iOS operating systems, and also available cross-platform through the iCloud website. iWork includes the presentation application Keynote, the word-processing and desktop-publishing application Pages, and the spreadsheet application Numbers. Apple's design goals in creating iWork have been to allow Mac users to easily create attractive documents and spreadsheets, making use of macOS's extensive font library, integrated spelling checker, sophisticated graphics APIs and its AppleScript automation framework. The equivalent Microsoft Office applications to Pages, Numbers, and Keynote are Word, Excel, and PowerPoint, respectively. Although Microsoft Office applications cannot open iWork documents, iWork applications can open Office documents for editing, and export documents from iWork's native formats (.pages, .numbers, .key) to Microsoft Office formats (.docx, .xlsx, .pptx, etc.) as well as to PDF files. The oldest application in iWork is Keynote, first released as a standalone application in 2003 for use by Steve Jobs in his presentations. Steve Jobs announced Keynote saying "It's for when your presentation really matters". Pages was released with the first iWork bundle in 2004; Numbers was added in 2007 with the release of iWork '08. The next release, iWork '09, also included beta access to iWork.com, an online service that allowed users to upload and share documents on the web, now integrated into Apple's iCloud service. A version of iWork for iOS was released in 2010 with the first iPad, and the apps have been regularly updated since, including the addition of iPhone support. In 2013, Apple launched iWork web apps in iCloud; even years later, however, their functionality is somewhat limited compared to equivalents on the desktop. iWork was initially sold as a suite for $79, then later at $19.99 per app on OS X and $9.99 per app on iOS. Apple announced in October 2013 that all iOS and OS X devices purchased onwards, whether new or refurbished, would be eligible for a free download of all three iWork apps: after device setup, the user can "claim" the apps on the App Store, after which they are permanently linked to the user’s Apple ID. iWork for iCloud, which also incorporates a document hosting service, is free to all iCloud users. iWork was released for free on macOS and iOS (including older or resold devices) in April 2017. In September 2016, Apple announced that the real-time collaboration feature would be available for all iWork apps. == History == The first version of iWork, iWork '05, was announced on January 11, 2005 at the Macworld Conference & Expo and made available on January 22 in the United States and on January 29 worldwide. iWork '05 comprised two applications: Keynote 2, a presentation creation program, and Pages, a word processor. iWork '05 was sold for US$79. A 30-day trial was also made available for download on Apple's website. Originally IGG Software held the rights to the name iWork. While iWork was billed by Apple as "a successor to AppleWorks", it does not replicate AppleWorks's database and drawing tools. However, iWork integrates with existing applications from Apple's iLife suite through the Media Browser, which allows users to drag and drop music from iTunes, movies from iMovie, and photos from iPhoto and Aperture directly into iWork documents. iWork '06 was released on January 10, 2006 and contained updated versions of both Keynote and Pages. Both programs were released as universal binaries for the first time, allowing them to run natively on both PowerPC processors and the Intel processors used in the new iMac desktop computers and MacBook Pro notebooks which had been announced on the same day as the new iWork suite. The next version of the suite, iWork '08, was announced and released on August 7, 2007 at a special media event at Apple's campus in Cupertino, California. iWork '08, like previous updates, contained updated versions of Keynote and Pages. A new spreadsheet application, Numbers, was also introduced. Numbers differed from other spreadsheet applications, including Microsoft Excel, in that it allowed users to create documents containing multiple spreadsheets on a flexible canvas using a number of built-in templates. iWork '09, was announced on January 6, 2009 and released the same day. It contains updated versions of all three applications in the suite. iWork '09 also included access to a beta version of the iWork.com service, which allowed users to share documents online until that service was decommissioned at the end of July 2012. Users of iWork '09 could upload a document directly from Pages, Keynote, or Numbers and invite others to view it online. Viewers could write notes and comments in the document, and download a copy in iWork, Microsoft Office, or PDF formats. iWork '09 was also released with the Mac App Store on January 6, 2011 at $19.99 per application, and received regular updates after this point, including links to iCloud and a high-DPI version designed to match Apple's MacBook Pro with Retina Display. On January 27, 2010, Apple announced iWork for iPad, to be available as three separate $9.99 applications from the App Store. This version has also received regular updates including a version for pocket iPhone and iPod Touch devices, and an update to take advantage of Retina Display devices and the larger screens of recent iPhones. On October 22, 2013, Apple announced an overhaul of the iWork software for both the Mac and iOS. Both suites were made available via the respective App Stores. The update is free for current iWork owners and was also made available free of charge for anyone purchasing an OS X or iOS device after October 1, 2013. Any user activating the newly free iWork apps on a qualifying device can download the same apps on another iOS or OS X device logged into the same App Store account. The new OS X versions have been criticized for losing features such as multiple selection, linked text boxes, bookmarks, 2-up page views, mail merge, searchable comments, ability to read/export RTF files, default zoom and page count, integration with AppleScript. Apple has provided a road-map for feature re-introduction, stating that it hopes to reintroduce some missing features within the next six months. As of April 1, 2014 a few features—e.g., the ability to set the default zoom—had been reintroduced, though scores had not. Due to using a completely new file format that can work across macOS, Windows, and in most web browsers by using the online iCloud web apps, versions of iWork beginning with iWork 13 and later do not open or allow editing of documents created in versions prior to iWork '09, with users who attempt to open older iWork files being given a pop-up in the new iWork 13 app versions telling them to use the previous iWork '09 (which users may or may not have on their machine) in order to open and edit such files. Accordingly, the current version for OS X (which was initially only compatible with OS X Mavericks 10.9 onwards) moves any previously installed iWork '09 apps to an iWork '09 folder on the users machine (in /Applications/iWork '09/), as a work-around to allow users continued use of the earlier suite in order to open and edit older iWork documents locally on their machine. In October 2015, Apple released an update to mitigate this issue, allowing users to open documents saved in iWork '06 and iWork '08 formats in the latest version of Pages. In 2016, Apple announced that the real-time collaboration feature would be available for all iWork apps, instead of being constrained to using iWork for iCloud. The feature is comparable to Google Docs. == Versions == === Major releases === === Updates === iWork '09 received several updates: iWork 9.0.3 DVD (for Mac OS X 10.5.6 "Leopard" or newer; released August 26, 2010) iWork 9.0.4 (for Mac OS X 10.5.6 "Leopard" or newer; released August 26, 2010) iWork 9.1 (for Mac OS X 10.6.6 "Snow Leopard" or newer; released July 20, 2011) iWork 9.3 (for Mac OS X 10.7.4 "Lion" or newer; released December 4, 2012) The Mac App Store version of iWork was updated on October 15, 2015 for 10.10 "Yosemite" or newer. It is the final release to support 10.10 "Yosemite" and 10.11 "El Capitan". Keynote 6.6, Pages 5.6 and Numbers 3.6 are included. iWork received a major update again on March 28, 2019 with Keynote 9.0, Pages 8.0 and Numbers 6.0. == Components == === Common components === Products in the iWork suite share a number of components, largely as a result of sharing underlying code from the Cocoa and similar shared application programming interfaces (APIs). Among these are the well known universal multilingual spell checker, which can also be found in products like Safari and Mail. Grammar checking, find and replace, style and color pickers are similar examples of design features found throughout the Apple application space. Moreover, the applications
Farthest-first traversal
In computational geometry, the farthest-first traversal of a compact metric space is a sequence of points in the space, where the first point is selected arbitrarily and each successive point is as far as possible from the set of previously-selected points. The same concept can also be applied to a finite set of geometric points, by restricting the selected points to belong to the set or equivalently by considering the finite metric space generated by these points. For a finite metric space or finite set of geometric points, the resulting sequence forms a permutation of the points, also known as the greedy permutation. Every prefix of a farthest-first traversal provides a set of points that is widely spaced and close to all remaining points. More precisely, no other set of equally many points can be spaced more than twice as widely, and no other set of equally many points can be less than half as far to its farthest remaining point. In part because of these properties, farthest-point traversals have many applications, including the approximation of the traveling salesman problem and the metric k-center problem. They may be constructed in polynomial time, or (for low-dimensional Euclidean spaces) approximated in near-linear time. == Definition and properties == A farthest-first traversal is a sequence of points in a compact metric space, with each point appearing at most once. If the space is finite, each point appears exactly once, and the traversal is a permutation of all of the points in the space. The first point of the sequence may be any point in the space. Each point p after the first must have the maximum possible distance to the set of points earlier than p in the sequence, where the distance from a point to a set is defined as the minimum of the pairwise distances to points in the set. A given space may have many different farthest-first traversals, depending both on the choice of the first point in the sequence (which may be any point in the space) and on ties for the maximum distance among later choices. Farthest-point traversals may be characterized by the following properties. Fix a number k, and consider the prefix formed by the first k points of the farthest-first traversal of any metric space. Let r be the distance between the final point of the prefix and the other points in the prefix. Then this subset has the following two properties: All pairs of the selected points are at distance at least r from each other, and All points of the metric space are at distance at most r from the subset. Conversely any sequence having these properties, for all choices of k, must be a farthest-first traversal. These are the two defining properties of a Delone set, so each prefix of the farthest-first traversal forms a Delone set. == Applications == Rosenkrantz, Stearns & Lewis (1977) used the farthest-first traversal to define the farthest-insertion heuristic for the travelling salesman problem. This heuristic finds approximate solutions to the travelling salesman problem by building up a tour on a subset of points, adding one point at a time to the tour in the ordering given by a farthest-first traversal. To add each point to the tour, one edge of the previous tour is broken and replaced by a pair of edges through the added point, in the cheapest possible way. Although Rosenkrantz et al. prove only a logarithmic approximation ratio for this method, they show that in practice it often works better than other insertion methods with better provable approximation ratios. Later, the same sequence of points was popularized by Gonzalez (1985), who used it as part of greedy approximation algorithms for two problems in clustering, in which the goal is to partition a set of points into k clusters. One of the two problems that Gonzalez solve in this way seeks to minimize the maximum diameter of a cluster, while the other, known as the metric k-center problem, seeks to minimize the maximum radius, the distance from a chosen central point of a cluster to the farthest point from it in the same cluster. For instance, the k-center problem can be used to model the placement of fire stations within a city, in order to ensure that every address within the city can be reached quickly by a fire truck. For both clustering problems, Gonzalez chooses a set of k cluster centers by selecting the first k points of a farthest-first traversal, and then creates clusters by assigning each input point to the nearest cluster center. If r is the distance from the set of k selected centers to the next point at position k + 1 in the traversal, then with this clustering every point is within distance r of its center and every cluster has diameter at most 2r. However, the subset of k centers together with the next point are all at distance at least r from each other, and any k-clustering would put some two of these points into a single cluster, with one of them at distance at least r/2 from its center and with diameter at least r. Thus, Gonzalez's heuristic gives an approximation ratio of 2 for both clustering problems. Gonzalez's heuristic was independently rediscovered for the metric k-center problem by Dyer & Frieze (1985), who applied it more generally to weighted k-center problems. Another paper on the k-center problem from the same time, Hochbaum & Shmoys (1985), achieves the same approximation ratio of 2, but its techniques are different. Nevertheless, Gonzalez's heuristic, and the name "farthest-first traversal", are often incorrectly attributed to Hochbaum and Shmoys. For both the min-max diameter clustering problem and the metric k-center problem, these approximations are optimal: the existence of a polynomial-time heuristic with any constant approximation ratio less than 2 would imply that P = NP. As well as for clustering, the farthest-first traversal can also be used in another type of facility location problem, the max-min facility dispersion problem, in which the goal is to choose the locations of k different facilities so that they are as far apart from each other as possible. More precisely, the goal in this problem is to choose k points from a given metric space or a given set of candidate points, in such a way as to maximize the minimum pairwise distance between the selected points. Again, this can be approximated by choosing the first k points of a farthest-first traversal. If r denotes the distance of the kth point from all previous points, then every point of the metric space or the candidate set is within distance r of the first k − 1 points. By the pigeonhole principle, some two points of the optimal solution (whatever it is) must both be within distance r of the same point among these first k − 1 chosen points, and (by the triangle inequality) within distance 2r of each other. Therefore, the heuristic solution given by the farthest-first traversal is within a factor of two of optimal. Other applications of the farthest-first traversal include color quantization (clustering the colors in an image to a smaller set of representative colors), progressive scanning of images (choosing an order to display the pixels of an image so that prefixes of the ordering produce good lower-resolution versions of the whole image rather than filling in the image from top to bottom), point selection in the probabilistic roadmap method for motion planning, simplification of point clouds, generating masks for halftone images, hierarchical clustering, finding the similarities between polygon meshes of similar surfaces, choosing diverse and high-value observation targets for underwater robot exploration, fault detection in sensor networks, modeling phylogenetic diversity, matching vehicles in a heterogenous fleet to customer delivery requests, uniform distribution of geodetic observatories on the Earth's surface or of other types of sensor network, generation of virtual point lights in the instant radiosity computer graphics rendering method, and geometric range searching data structures. == Algorithms == === Greedy exact algorithm === The farthest-first traversal of a finite point set may be computed by a greedy algorithm that maintains the distance of each point from the previously selected points, performing the following steps: Initialize the sequence of selected points to the empty sequence, and the distances of each point to the selected points to infinity. While not all points have been selected, repeat the following steps: Scan the list of not-yet-selected points to find a point p that has the maximum distance from the selected points. Remove p from the not-yet-selected points and add it to the end of the sequence of selected points. For each remaining not-yet-selected point q, replace the distance stored for q by the minimum of its old value and the distance from p to q. For a set of n points, this algorithm takes O(n2) steps and O(n2) distance computations. === Approximations === A faster approximation algorithm, given by Har-Peled & Mendel (2006), applie