skos PM RelatedTo seeAlso webprotege:R72QfF3SXvFG37XoRGvdcxh Complexity in Adaptive Systems Active Control [NASA p.10] Adaptive Control [TRT] External Complexity [Natural Language, 2021] Internal Complexity [Natural Language, 2021] An adaptive system, or complex adaptive system, is a special case of complex systems, which is able to adapt its behavior according to changes in its environment or in parts of the system itself. In this way, the system can improve its performance through a continuing interaction with its environment. The concept of complexity in an adaptive system is used to analyze the interactive relationship between the system and its environment, which can be classified into two types: internal complexity for model complexity and external complexity for data complexity. The internal complexity is defined by the amount of input, information, or energy that the system receives from its environment. The external complexity refers to the complexity of how the system represents these inputs through its internal process (He, Jun, 2020) Adaptive Systems [INSPEC] webprotege:R7KBAUGP2gkMUa2zSTWTur Clustering from Data Streams Cluster Analysis [NASA-p.171] Clustering [Natural Language, 2021] Data Streams Grouping [Natural Language, 2021] Clustering is the process of grouping objects into different groups, such that the common properties of data in each subset are high and between different subsets are low. The data stream clustering problem is defined as to maintain a continuously consistent good clustering of the sequence observed so far, using a small amount of memory and time. The issues are imposed by the continuous arriving data points and the need to analyze them in real time. These characteristics require incremental clustering, maintaining cluster structures that evolve over time. Moreover, the data stream may evolve over time, and new clusters might appear, other disappears, reflecting the dynamics of the stream (Gama, Joao, 2020). The analysis of data with the object of finding natural groupings within the data either by hand or with the aid of a computer. [NASA] Pattern Clustering [INSPEC] webprotege:R7elMSxMscPrsWYpPbDWGFN Deep Learning NN [Natural Language, 2021] Neural Nets [NASA-642] Neural Networks [Natural Language, 2021] Deep learning Deep learning has revolutionized Pattern Recognition and Machine Learning. It is about credit assignment in adaptive systems with long chains of potentially causal links between actions and consequences… A standard NN consists of many simple, connected processors called units, each producing a sequence of real-valued activations. Input units get activated through sensors perceiving the environment, other units through connections with real-valued weights from previously active units. Some units may influence the environment by triggering actions. Learning or credit assignment is about finding weights that make the NN exhibit desired behavior, such as controlling a robot. Depending on the problem and how the units are connected, such behavior may require long causal chains of computational stages, where each stage transforms (often in a nonlinear way) the aggregate activation of the network. Deep learning in NNs is about accurately assigning credit across many such stages (Schmidhuber, Jurgen, 2020). Neural Nets [INSPEC] webprotege:R7jWJty3P5NZMnLJu9UPYn9 Feature Selection Dimensionality Reduction Technique [Natural Language, 2021] Feature Identification and Location Experiment [NASA-p.338] Feature selection, as a dimensionality reduction technique, aims to choose a small subset of the relevant features from the original ones by removing irrelevant, redundant, or noisy features. Feature selection usually leads to better learning performance, i.e., higher learning accuracy, lower computational cost, and better model interpretability (Wang, Suhang, Tang, Jiliang, and Liu, Huan, 2020). Feature Extraction [INSPEC] webprotege:R8xox6Q2uEFkbZRQtoT8Bcz Regression Parametric Functions [Natural Language, 2021] Regression Analysis [NASA-p.814] Regression Analysis [TRT] Regression Function [Natural Language, 2021] The statistical counterpart or analog of the functional expression, in ordinary mathematics, of one variable in terms of others. [NASA] Regression Analysis [INSPEC] webprotege:RBLuHo6LnrfCb0Wk3icZ9DA Weight Weight [NASA-p.1074] Weighted Connections [Natural Language, 2021] Weighted Networks [Natural Language, 2021] Weighted Neurons [Natural Language, 2021] Weighting [TRT] The force exerted on a body by gravity [NASA]. Weight [ML Encyclopedia] webprotege:RBS3Y0j5zHkSAL3tUbs0cUz Computational Complexity of Learning Complexity Theory [Natural Language, 2021] Computational Complexity [Natural Language, 2021] Measures of the complexity of learning have been developed for a number of purposes including Inductive Inference, PAC Learning, and Query-Based Learning. The complexity is usually measured by the largest possible usage of resources that can occur during the learning of a member of a class. Depending on the context, one measures the complexity of learning either by a single number/ordinal for the whole class or by a function in a parameter n describing the complexity of the target to be learned. The actual measure can be the number of mind changes, the number of queries submitted to a teacher, the number of wrong conjectures issued, the number of errors made, or the number of examples processed until learning succeeds. In addition to this, one can equip the learner with an oracle and determine the complexity of the oracle needed to perform the learning process. Alternatively, in complexity theory, instead of asking for an NP-complete oracle to learn a certain class, the result can also be turned into the form “this class is unlearnable unless RP=NP” or something similar. (Here RP is the class of decision problems solvable by a randomized polynomial time algorithm, and NP is the class of decision problems solvable by a nondeterministic polynomial time algorithm, and both algorithms never give “yes” answer for an instance of the problem with “no” answer.) (Jain, Sanjay and Stephan, Frank, 2020) Computational Complexity [INSPEC] webprotege:RBViBDAKSVULibhBIXD2uHx Precision and Recall Confusion Matrix [Natural Language, 2021] Information Retrieval [Natural Language, 2021] Precision and recall are the measures used in the information retrieval domain to measure how well an information retrieval system retrieves the relevant documents requested by a user. The measures are defined as follows: Precision=Total number of documents retrieved that are relevant/total number of documents that are retrieved Recall=Total number of documents retrieved that are relevant/total number of relevant documents in the database We can employ the same terminology used in a confusion matrix to define these two measures. Let relevant documents be positive examples and irrelevant documents, negative examples. The two measures can be redefined with reference to a special case of the confusion matrix, with two classes: one designated the positive class and the other the negative class. Precision=True positives/total number of positives predicted=TP/(TP+FP). Recall=True positives/total number of actual positives DTP/(TP+FN). Instead of two measures, they are often combined to provide a single measure of retrieval performance called the F-measure as follows: F-measure = 2 * recall * precision/(recall + precision) (Ming Ting, Kai, 2020). Precision and Recall [ML Encyclopedia] webprotege:RCEF5OjKtQIblIAKVzikMJ Classifier Calibration Classifier Rules [Natural Language, 2021] Classifier [Natural Language, 2021] Linear Classifier [Natural Language, 2021] Classifier calibration is concerned with the scale on which a classifier’s scores are expressed. While a classifier ultimately maps instances to discrete classes, it is often beneficial to decompose this mapping into a scoring classifier which outputs one or more real-valued numbers and a decision rule which converts these numbers into predicted classes. For example, a linear classifier might output a positive or negative score whose magnitude is proportional to the distance between the instance and the decision boundary, in which case the decision rule would be a simple threshold on that score. The advantage of calibrating these scores to a known, domain independent scale is that the decision rule then also takes a domain-independent form and does not have to be learned. The best-known example of this occurs when the classifier’s scores approximate, in a precise sense, the posterior probability over the classes; the main advantage of this is that the optimal decision rule is to predict the class that minimizes expected cost averaged over all possible true classes. The main methods to obtain calibrated scores are logistic calibration, which is a parametric method that assumes that the distances on either side of the decision boundary are normally distributed and a nonparametric alternative that is variously known as isotonic regression, the pool adjacent violators (PAV) method or the ROC convex hull (ROCCH) method (Flatch, Peter, 2020). Classifier Calibration [ML Encyclopedia] webprotege:RCU7OVlaTt2xgNzHCdlvbH0 Topology of Neural Networks NN [Natural Language, 2021] Neural Network [Natural Language, 2021] Topology Network [Natural Language, 2021] Topology [NASA-p.1005] Topology [TRT] Topology [INSPEC] webprotege:RCyJsxUAqZpMr2cdUnfLT0u Classification Rule Boolean Rules [Natural Language, 2021] Classifications [NASA-p.167] IF-THEN Rules [Natural Language, 2021] A classification rule is an IF-THEN rule. The condition of the rule (the rule body or antecedent) typically consists of a conjunction of Boolean terms, each one constituting a constraint that needs to be satisfied by an example. If all constraints are satisfied, the rule is said to fire, and the example is said to be covered by the rule. The rule head (also called the consequent or conclusion) consists of a single class value, which is predicted in case the rule fires. This is in contrast to association rules, which allow multiple features in the head. (Fürnkranz, Johannes, 2020) Classifications [INSPEC] webprotege:RXp4U3gsuesc7yUnwmqJY4 Evaluation of Learning Algorithms Algorithm Properties [Natural Language, 2021] Algorithm Suitability [Natural Language, 2021] Backpropagation (artificial intelligence) [NASA-p.87] Learning Algorithms [Natural Language, 2021] Learning (artificial intelligence) [INSPEC] owl:Thing Subset mapping Boolean