Week One Terms Flashcards
Study Up On DS Terminology
Define Actor-critic
A reinforcement learning algorithm that simultaneously learns a policy function and a value function to take the best from both.
Define Adjoint
For a finite-dimensional linear map (I.e: a matrix A) the adjoint A* is given by the complex conjugate transpose of the matrix. In the infinite-dimensional context, the adjoint A* of a linear operator A is defined so that (Af,g) = (f,A*g), (where < .,.> is an inner product.
Define Agent
In reinforcement learning (RL), an agent senses the state (s) of its environment and learns to take appropriate actions (a) to achieve an optimal future reward (r).
Define Akaike information criterion (AIC)
An estimator of the relative quality of statistical models for a given data set. Given a collection of models for the data, AIC estimates the quality of each model relative to each of the other models. Thus, AIC provides a means for model selection.
Define Autoencoder
Autoencoders are machine-learning models that learn efficient latent coding of unlabeled data (unsupervised learning). Autoencoders learn efficient codings by performing nonlinear dimensionality
reduction. Autoencoders are typically trained with both an encoding layer and
a decoding layer so that one can map to the latent representation and back.
Define Backpropagation (backdrop)
A method used for computing the gradient descent required for the training of neural networks (NNs). Based upon the chain rule, backdrop exports the compositional nature of NNs in order to frame in optimization problem for updating the weights of the net work. It is commonly used to train deep neural networks (DNNs).
Define Balanced input-output model
A model expressed in a coordinate system, where the states are ordered hierarchically in terms of their joint controllability in observability. The controllability and observability Gramians are equal and diagonal for such a system.
Define Bayesian Information Criterion (BIC)
An estimator of the relative quality, a statistical models for a given set of data. Given a collection of models for the data, DIC estimates the quality of each model, relative to each of the other models. Thus, BIC provides a Memes for model selection.
Define Bellman Optimality
A cornerstone of dynamic programming, stating that an optimal multi step sequence, must also be locally optimal in every sub sequence of steps.
Define Classification
A general process related to categorization, the process, in which ideas and objects are recognized, differentiated, and understood. Classification is a common task for machine learning algorithms.
Define Closed-loop Control
A control architecture where the actuation is informed by sensor data about the output of the system.
Define Clustering
A task of grouping, a set of objects in such a way that objects in the same group (called a cluster) are more similar (in some sense) to each other than to those in other groups (clusters). It is a primary goal of exploratory data, mining, and a common technique for statistical data analysis.
Define Coherent Structure
A spatial mode that is correlated with the data from the system.
Define Compressed Sensing
The process of re-constructing a high dimensional vector signal from a random under sampling of the data, using the fact that the high dimensional signal is sparse in an unknown transform basis, such as the Fourier basis.
Define Compression
The process of reducing the size of a high dimensional, vector or array by proxy meeting it as a sparse vector in a transform basis. For example, MP3 and JPG compression use the Fourier bases or wavelet basis to compress, audio, or image signals.
