Dave's PCF WIP: Paragraphs

Paragraphs

Actions	Application	Content	Paragraph Number	Notes	Modified
		Content		Paragraph Number	Notes	Any Field
View Edit Delete	PER-17	Developmental learning can include robotic learning, which can generate its own learning situations to acquire new skills through autonomous self-exploration and interaction, for example, with human teachers.	76	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	A goal of machine learning or deep learning can be to generalize from the experience. Generalization can be the ability of a learning machine or system to perform accurately on new, unseen inputs after having been trained on a training data set. Training examples can come from an unknown probability distribution and the learning machine or system can be tasked to build an input and output model that enables it to produce sufficiently accurate predictions with new inputs. Bounds or limits, e.g. probabilitistic bounds, on the performance, accuracy and reproducibility of machine learning can be determined. When machine learning is used for solving clinical problems, e.g. outcome prediction or treatment planning or modification, the bounds or limits of performance, accuracy and/or reproducibility of the machine learning system or learning machine can influence and/or determine the performance, accuracy, and/or reproducibility of the clinical application, e.g. outcome prediction or treatment planning or modification. Accuracy can include the assessment of true positives, true negatives, false positives and false negatives. Reproducibility can be precision. Performance of the machine learning system and/or learning machine can include other statistical measures known in the art for assessing the performance of a clinical system.	77	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Learning systems including machine learning can use decision tree learning, association rule learning, artificial neural networks (ANNs), deep learning, inductive logic programming, support vector machines, clustering, Bayesian networks, reinforcement learning, representation learning, similarity and metric learning, sparse dictionary learning, biologic or genetic algorithms, rule based machine learning and learning classifier systems.	78	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	With decision tree learning, a decision tree can be used as a predictive model, which can map observations about one or more parameters to conclusions about the parameter's target value. With association learning, relations between variables or parameters can be identified in large databases. With ANNs, computations can be structured through an interconnected group of artificial neurons, processing information using a connected approach. ANNs can be non-linear data modeling took, using various statistical methods and approaches known in the art. Deep learning can employ multiple layers in an artificial neural network. Inductive logic programming (ILP) can utilize logic programming for rule learning, e.g. using a uniform representation for input examples, background knowledge, and hypotheses. Support vector machines (SVMs) can be a set of supervised learning methods used for classification and/or regression. A given set of training examples can be marked as belonging to a first category or a second category; an SVM training machine can build a model predicting whether a new input falls into the first or the second category.	79	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Clustering can be the assignment of a set of observations into subsets, where data in each subset have similarities with regard to one or more parameters while data in different subsets can be dissimilar with regard to one or more parameters. Clustering techniques can provide information on similarity or dissimilarity, for example reflected in a similarity metric, a measurement of internal compactness or separation between different clusters. A Bayesian network can be a graphical model representing, for example, random variables and their conditional independencies. This can be shown in a directed acyclic graph. A Bayesian network can represent the probabilistic relationships between diseases and symptoms. A Bayesian network can be used to compute the probability(ies) of the presence of one or more disease.	80	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	With reinforcement learning, input and output pairs can never be presented; reinforcement learning can map a state, e.g. a clinical state of a patient, and can develop predictions, actions or treatment the system can make. With representation learning algorithms input information can be preserved but transformed to make it more useful, e.g. as a pre-processing step, prior to performing classification or predictions, allowing reconstruction of the inputs coming from the unknown data generating distribution.	81	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Deep learning can utilize multiple levels of representation, e.g. in an ANN. Higher-level, e.g. more abstract, parameters or data can be defined as generating lower-level parameters or data. With similarity learning, the learning system or machine can be given pairs of data that are considered similar and pairs of less similar data. It can then learn a similarity function or a distance metric function that can predict if new objects are similar.	82	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Rule-based machine can be identification and utilization of a set of relational rules that can represent the knowledge captured by the learning system. Learning classifier systems (LCS) can be a family of rule based machine learning algorithms or systems which can combine a discovery component with a learning component.	83	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	The accuracy of classification machine learning models can be evaluated using accuracy estimation techniques and statistical techniques and methods testing the accuracy, sensitivity, specificity, fake positive and false negative rates. Other statistical methods such as Receiver Operating Characteristic (ROC) and associated Area under the Curve (AUC) as well as Total Operating Characteristic (TOC) can be used. (See US Patent 11,278,413 Para. 64-75).	84	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Deep Learning	85	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Deep learning can include machine learning algorithms which can use multiple layers of non-linear processing units or elements. Each layer can use the output from a higher layer as input. Deep learning systems can work in a supervised setting, e.g. using one more classification systems. Deep learning systems can also work in an unsupervised setting, e.g. in the context of texture analysis or pattern recognition. Deep learning systems can learn multiple levels of representations that correspond to different levels of abstraction. The different levels can form an order or a hierarchy of concepts. The different layers of a deep learning system can reside in different layers of an artificial neural network, i.e. a deep neural network. They can include hidden layers in an ANN. Deep learning systems and deep ANNs can utilize Boltzmann machines. With deep learning systems, layers can correspond to layers of abstraction, e.g. across a deep neural network. Varying numbers of layers and layer sizes can provide different degrees of abstraction. Higher level, more complex concepts can be learned from lower level layers.	86	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Deep neural networks (DNNs) can be one or more ANNs with multiple hidden layers between the input and output layers. DNNs can model complex non-linear relationships. DNNs can generate models where the object is expressed as a layered composition. DNNs can be feedforward networks in which data flows from the input layer to the output layer without looping back, DNNs can be recurrent neural networks or convolutional deep neural networks.	87	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Deep learning algorithms can be applied to unsupervised learning tasks. This is an important benefit because unlabeled data can be more abundant than labeled data. For example, in a clinical environment, a deep learning system with a multi-layered ANN can initially be trained using a classification of outcomes in a supervised fashion. As the data grows, the system can optionally learn in an unsupervised manner, for example by utilizing pattern recognition across large clinical datasets, which can include pre-operative, intra-operative and post-operative data. (See US Patent 11,278,413 Para. 76-79).	88	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Classification	89	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Classification can be a process of creating categories, in which data or objects can be recognized, differentiated or understood. A classification system can be an approach of accomplishing classification. Classification can be performed using mathematical classification, statistical classification, classification theorems, e.g. in mathematics, and attribute value systems. Classifications can be alphanumeric. Classifications can be single or multi-dimensional. Classifications can be single or multi-layered. Classifications can be color coded. An ANN can use a single classification system, e.g. in supervised learning. An ANN can use multiple classification systems. When multiple classification systems are used, they can reside in different layers of a DNN or deep learning system.	90	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Classification can be the problem of identifying to which of a set of categories or sub-populations a new observation belongs; this can be determined, for example, using a training data set with observations whose category membership is known. For example, a diagnosis can be assigned to a patient as a category which can be described by measured data or characteristics such a heart rate, blood pressure, presence of absence of symptoms or combinations of symptoms. Classification can be a pattern recognition.	91	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Individual observations or data can be divided into a set of quantifiable properties. These properties may be categorical, e.g. “a”, “b”, “c”, “d” etc. or ordinal, e.g. “excellent”, “very good”, “good”, “acceptable”, “average”, or “poor”. They can be integer or real valued. Observations or data can also be classified using similarity or distance functions, e.g. based on earlier observations or data. An algorithm that implements classification can be a classifier, A classifier can sometimes also be a mathematical function, e.g. implemented by a classification algorithm, that can map input data to a category.	92	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Data can be classified, for example, into “excellent”, “good”, “acceptable” or “poor” outcome, e.g. one or more clinical outcomes or clinical outcome variables. Numeric values or ranges of numeric values can be assigned to different classes, for example numeric values or ranges of numeric values from a patient reported outcome measurement, from a clinical reporting system, and/or from one or more electronic measurements.	93	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	Discriminative vs. Generative Models and Networks	94	Added by DJM Jan 2024	1/6/24, 10:05 PM
View Edit Delete	PER-17	In machine learning, discriminative models can be distinguished from generative models.	95	Added by DJM Jan 2024	1/6/24, 10:05 PM

Page 14 of 20, showing 20 record(s) out of 389 total