Concepts starting with A

225 concepts, each with a plain-language explanation, examples and diagrams. Pick a letter or open a concept.

A Clustering Model — The supplied source pack does not define PCA or variance maximization.
A First Look at a Neural Network — Advanced deep learning combines tensor-based data representation, tensor operations, and gradient-based optimization.
A Trivial Client/Server Example — DateServer creates a ServerSocket on LISTENING_PORT, announces that it is listening, and then loops on accept(), handling one…
A Year in the Life of a Forward Deployed Engineer — Onboarding for a Forward Deployed Engineer combines meetings with senior engineers, reading the team's playbook documents, and…
Absolute Positional Embeddings — A token embedding is a vector representation of an input token, created in PyTorch with an Embedding layer.
Abstract Base Classes and Interfaces — Polymorphism allows a subtype to be treated as an instance of its parent type, enabling flexible code that works with multiple…
Abstract Classes — An abstract class is declared with the abstract modifier and is meant to be subclassed rather than instantiated on its own.
Abstract Classes and Methods — A Java interface is a reserved word that introduces a set of instance method declarations with no implementations.
Abstract Classes Versus Interfaces — A default method is a method declared in an interface with a body, so implementing classes inherit a shared implementation…
Abstract Methods — An abstract class is declared with the abstract modifier and is meant to be subclassed rather than instantiated on its own.
Abstract Methods and Interfaces — A default method is declared inside an interface with the modifier default and carries a working implementation.
Abstract Methods in Java — A default method is a method with an implementation in an interface, marked with the default keyword, and classes that implement…
Abstraction of low-level details — The characteristic of high-level languages where the programmer does not need to manage low-level system details.
Access Control and Getter/Setter Methods — A getter method provides read access to a variable by returning its current value to the caller.
Access Control for Agents — A complete audit log is not sufficient for compliance if its entries cannot identify the human responsible for each read.
Access Reviews and Renewal Cycles — Every agent needs its own identity, and permissions, expiration, and audit records all attach to that identity.
Accessing and Modifying Array Elements — An array type is the base type followed by a pair of empty square brackets, such as int[] or String[].
Accessing Array Elements by Index — An array can be initialized by listing its values, separated by commas, between a pair of braces, and its length is the number…
Accessing Attributes and Methods — An object bundles data (attributes) and behavior (methods) into a single, organized unit.
Accessing Dictionary Values — A dictionary is a mapping between keys and values, not a sequence indexed by position like a list.
Accessing Dictionary Values with the Index Operator — A for loop over a dictionary iterates through keys; to filter by value, use the index operator to retrieve each value and test…
Accessing list elements with indexing — A list is a sequence of values that can contain any data type, unlike strings which contain only characters.
Accessing Object Attributes and Methods — The real power of object-oriented programming is creating multiple instances of a class, each with its own independent data.
Accessing Single Elements with find() — findall() returns a Python list of all Element objects matching a given path in the XML tree.
Accessing Support Resources — The forum is a venue for meaningful reader discussion and reader-author dialogue.
Accessing the Source Code — The book's code examples are available for download.
Accessing XML Attributes and Elements — The findall method retrieves a list of XML nodes using a path that must include all parent-level elements (except the root)…
Account Ownership — Successful FDEs treat themselves as the CEO and CTO of their own account, which makes four claims of ownership: the account, the…
Accountability in Customer Interactions — Simulation interviews ask you to recount specific past customer engagements, so specificity and clear structure matter more than…
Accumulation Patterns in Programming — The counter pattern is a fundamental computational technique: initialize a counter to 0, loop through data, check a condition…
Accumulator Patterns with Strings — Traversal is the process of handling a string one character at a time, from beginning to end.
Accumulators and Running Totals — A two-dimensional array is traversed with nested loops: the outer loop chooses a row and the inner loop visits each column of…
Accuracy and Confidence in PAC Learning — Sample complexity gives the number of examples required for a probably approximately correct solution.
Achieving Human-Level Performance in Video Games with Deep Reinforcement Learning — Experience replay and the duplicate target network each improved DQN, and their combination produced a very dramatic performance…
Action Events and Listeners — Event-oriented programming lets classes communicate by having one class emit events that other classes listen for and respond to.
Action Preferences and Softmax Action Probabilities — The gradient bandit algorithm can be understood as a stochastic approximation to gradient ascent.
Action Selection — Dyna-Q+ is Dyna-Q with an exploration bonus.
Action Selection from the Current State — A state signal may combine immediate sensations with information derived from past sensations.
Action Selection in Bandit Problems — q∗(a) denotes the true value of action a.
Action Selection in Reinforcement Learning — A k-Armed Bandit Problem is a repeated decision problem with k possible actions.
Action Selection Methods — Semi-gradient Sarsa with function approximation was first explored by Rummery and Niranjan in 1994.
Action Value Computation — Action value maximization can be dangerous when an incorrect response would severely harm Watson's chances of winning.
Action Values and Policies in Reinforcement Learning — Model-free decisions use cached values or a cached policy rather than an environment model.
Action Values in Policy-Gradient Methods — A baseline extends the policy gradient theorem by providing a reference for comparing an action value.
Action-Value Estimates — Greedy actions are those with the greatest current estimated value.
Action-Value Estimates and Step-Size Parameters — High initial action-value estimates can make a greedy method explore without explicitly using ε-greedy random selection.
Action-Value Estimation — UCB combines an estimated action value with an uncertainty bonus.
Action-Value Functions — The optimal state-value function v∗(s) gives the maximum expected return achievable from a state.
Action-Value Functions in Scheduling Agents — Sarsa supplies the learning method for the DRAM scheduling agent.
Action-Value Methods — A sample average is a natural estimate of an action value.
Action-Value Methods for Action Selection — Numerical preferences express how strongly each action is favored.
Action-Value Methods for Decision Making — Greedy action selection chooses an action with the highest estimated action value.
Action-Value Methods in Reinforcement Learning — n-step Q(σ) is a unifying framework for action-value backups.
Action-value update rule — The n-step target is a quantity used by an action-value update rule.
ActionListener and Event Handling — A javax.swing.Timer exists to generate ActionEvent events at a fixed delay, and it can also be set to fire a single event as an…
Actions and Environmental Responses — An agent learns and makes decisions through interaction.
Actions, States, and Rewards in Reinforcement Learning — A reward signal helps define the reinforcement learning problem together with the environment.
Activation Functions — A feedforward neural network is a directed acyclic graph whose nodes are neurons and whose directed edges are weighted…
Activation Functions in Deep Neural Networks — DeepDream changes an existing image by maximizing activations inside a convolutional network.
Actor and Critic Learning Rules Explained — An eligibility trace is a temporary synaptic record that lets later reinforcement assign credit to earlier activity.
Actor-Critic Algorithms — Non-contingent eligibility traces do not correlate actions with consequent changes in the reward signal.
Actor-Critic Brain Implementation — In the hypothetical actor-critic brain implementation, the ventral striatum sends value information to the VTA and SNpc.
Actor-Critic Learning Rules — An actor-critic neural network contains distinct critic and actor components.
Actor-Critic Reinforcement Learning — The actor-critic framework can be connected to synaptic plasticity rather than treated only as an abstract algorithm.
AdaBoost — Model selection is the task of choosing an algorithm for a particular problem and setting its parameters.
AdaBoost Algorithm — Model selection combines algorithm choice with parameter setting.
AdaBoost and Distribution-Weighted Risk — Decision stumps are threshold-based hypotheses of the form sign(θ - x_i).
Adapting Product Decisions Without Losing Direction — Founder-mode temperament is a mindset that holds a core thesis tightly while staying loose on the decisions downstream of it.
Adaptive Scheduling Policy — On-line learning adapts during system operation.
Adding Actions to Buttons in Java GUI Programming — A JToolBar is created with new JToolBar() and placed in a container, such as along BorderLayout.NORTH, using an add call with a…
Adding Tensors — Broadcasting makes tensors with different shapes compatible for operations such as addition when possible and unambiguous.
Addition and Division Operators — Subtraction removes the right operand from the left; unary minus (a single minus with no left operand) negates a number.
Additional Project Showcase Elements — The eval set is documented in /evals, with a README inside that directory.
Addressing Information in Internet Communication — Packets are the units used for communication over the Internet.
Adjusting the Tokenizer for Special Tokens — Special tokens such as '<|endoftext|>' and '<|unk|>' are added to the token list after the regular unique words, using extend().
Advanced Deep Learning Techniques — Chapter 5 examines practical computer-vision examples with a focus on image classification.
Advanced deep-learning best practices — Deep learning became more popular through the combined influence of hardware, data, algorithms, investment, and democratization.
Advanced List Comprehension with Conditions — List comprehension creates a new sequence from an existing one by applying a transformation to each item in a single, compact…
Advanced Regex: Lookahead and Lookbehind Assertions — re.findall() is a Python method that extracts all non-overlapping substrings matching a regex pattern from a string.
Advanced Set Operations — Sets are unordered collections of unique elements, used when membership matters more than order or frequency.
Advanced Techniques for Deep Learning Models — The Functional API describes a model as a connected route of tensor transformations.
Advanced Unpacking Techniques — A function can return multiple values by packaging them into a single tuple object using the comma syntax in the return statement.
Advanced Use of Recurrent Neural Networks — The supplied section places bidirectional recurrent layers within advanced techniques for recurrent neural networks.
Advantages of Temporal Difference Prediction Methods — Bootstrapping is the process of learning a guess from another guess.
Advantages Over Dynamic Programming Methods — TD methods learn transition by transition, whereas Monte Carlo methods wait until an episode ends.
Adversarial Environments in Online Learning — A hypothesis class with only two constant predictors is still vulnerable to an adversary that reacts to the learner's current…
Advisory Roles in AI Engineering — A Forward Deployed Advisor is a senior practitioner who runs a solo consultancy and sells FDE work without performing it.
Afterstate Methods — Afterstate methods preserve the generalized policy iteration perspective.
Afterstate Value — Action value computation evaluates a candidate bet by its expected resulting value.
Agent-Environment Boundary and Policy — Locate the boundary by asking what the agent can change arbitrarily, not by tracing the body's outer surface.
Agent-Environment Interaction Basics — The agent-environment interaction is a repeating sequence of state, action, reward, and next state.
Agentic Tools in the Delta's Workflow — The autonomy slider sets how far up the Delta is pushing the next generation of agentic tools, and the governing question is how…
Aggregate Functions and GROUP BY Clauses — JOIN and ON clauses reconstruct data across multiple tables by following relational connections between foreign keys and primary…
Aggregate Functions: Summarizing Data with COUNT, SUM, AVG — SQL uses a single equal sign (=) for equality testing in WHERE clauses, not the double equal sign (==) used in Python.
Agnostic PAC Learnability — The upper bound shows that a sufficiently large sample can make ERM an agnostic PAC learner.
Agnostic PAC Model — Regret measures how much worse an online algorithm performed than a fixed reference predictor.
AI Agent Impact on Pricing — Per-seat pricing charges enterprise customers based on the number of human users, making headcount the anchor of the price.
AI System Governance and Compliance — SOC2 is a control framework that is audited annually, and customers rely on it when winning enterprise contracts.
Algorithm Development — An algorithm is a clear sequence of steps for carrying out a task.
Algorithm Running Time Analysis — Big-Oh notation describes an upper bound on how an algorithm's running time grows as the problem size n increases without limit.
Aliasing and References in Python — Lists are mutable and the default choice for most tasks; tuples are immutable and solve specific problems.
Aligning AI Projects with Organizational Priorities — The first filter for an AI project is whether the problem is one of the customer's top five priorities; the source says that if…
All-Pairs Reduction — General multiclass-to-binary reductions build multiclass predictors from several binary classifiers.
All-Pairs Reduction for Multiclass Categorization — One-versus-All represents a multiclass hypothesis using one binary hypothesis for each label.
AlphaGo's Policy and Value Networks — AlphaGo evaluated game states with both a value network and rollouts.
AlphaGo's Rollout Policy — AlphaGo evaluated game states with both a value network and rollouts.
AlphaGo’s Supervised-Learning Policy Network — The reinforcement-learning policy network began with the final weights of the supervised-learning policy network.
AlphaGo's Value Network — AlphaGo evaluated game states with both a value network and rollouts.
Alternative Data Formats: JSON and CSV — Data persistence allows you to store Python objects in files so they survive program termination.
Analysis of 1-NN Rule — Higher dimension creates an exponential, not merely proportional, demand for training data in this analysis.
Analyzing Algorithm Efficiency — Asymptotic analysis studies how an algorithm's run time depends on problem size as that size increases without limit.
Analyzing Claims Processing Workflows — A bottleneck is a part of a process that limits the overall performance, so it is defined by its effect on the whole flow.
Analyzing Communication Patterns in Email Networks — Raw email data contains inconsistencies (mixed-case addresses, subdomain variations, gateway artifacts) that must be…
Analyzing Contributor Networks and Participation Patterns — Email addresses and domain names change over time as people move between organizations, requiring a mapping system to…
Analyzing Contributor Networks with Unified Identities — Mapping tables unify changing email addresses and multiple domain names into single canonical identifiers, preventing entity…
Analyzing Network Connectivity Patterns — Visualizing a network graph requires running spjson.py to convert your PageRank database into JSON data, then opening force.html…
Analyzing Web Connectivity with PageRank — spider.py is the script used to crawl websites and record links.
Anatomy of a Neural Network — Tensors provide the data representations used by neural networks, from 0D scalars through higher-dimensional forms.
Anchoring at the Beginning: Using ^ to Match Line Starts — The $ anchor marks the end of a line in regular expressions, not a literal dollar sign character
Anchors and Boundaries in Regex — String splitting for data extraction is brittle and requires extensive error handling; regex extraction is more flexible and…
Anchors and Boundaries in Regular Expressions — Square brackets [...] define a set of acceptable characters for a single position in a regex pattern.
Anchors and Word Boundaries for Precise Matching — Greedy quantifiers expand to match the last occurrence of a pattern; non-greedy quantifiers match the first occurrence.
Anchors: Matching at the Start or End of a Line — The period (.) in a regular expression is a wildcard that matches any single character—letters, digits, spaces, punctuation, or…
Animal Training — Shaping changes the reward rule gradually rather than demanding the final behavior at once.
Annual Contract Value and Revenue Measurement — The contract-size floor is the minimum annual contract value below which the FDE motion is not viable, placed by the source at…
Anonymous Classes — A lambda expression is a shorthand for a function, and it can only be used with a functional interface, meaning an interface…
Anonymous Event Handlers and Adapter Classes — An anonymous inner class is a class defined without a name, and a single new expression both declares it and creates an object…
Anonymous Functions in Python — A lambda statement creates an anonymous function object with one parameter and one expression.
Anthropic Plugin Development — A plugin is a self-contained package of prompts, tools, and configuration for one specific task.
API Design Best Practices — Service-oriented architecture enables travel websites to integrate multiple independent systems (airlines, hotels, car rentals)…
API Keys and Basic Authentication — Vendors require cryptographically signed messages to verify both the source of requests and that they have not been altered in…
APIs and Data Exchange — JSON (JavaScript Object Notation) was inspired by JavaScript but influenced by Python's dictionary and list syntax, making it…
APIs, Packages, and Modules in Java — A specific import names one class using its fully qualified name, in the form import package_name.ClassName;
Appending Data to Files: Mode 'a' — Write mode ('w') opens a file for writing and immediately truncates (erases) any existing content in the file.
Applied Machine Learning Skill Set — The mental model for the FDE role is closer to statistical engineering than to classical software engineering.
Applying Deep Learning in Real-World Scenarios — Fine-tuning starts with a pre-trained convnet rather than an uninitialized model.
Applying Frameworks to Understand Customer Needs — Continuous Discovery Habits is a discipline of regular customer interviews that inform product development decisions.
Approximate Solution Methods — Large state spaces create problems of memory, time, and data, and many encountered states may never have appeared before.
Approximate Solution Methods in Reinforcement Learning — A tabular method stores value-function estimates directly in arrays or tables.
Approximate Solutions for Large State Spaces — Function approximation turns examples of a desired function into a broader estimate.
Approximate Solutions in Reinforcement Learning — The Bellman optimality equation gives a route to finding an optimal policy, but explicit solution is rarely directly useful.
Approximate Value Functions — SGD updates can be used with linear function approximation.
Approximation Error — A hypothesis class is the set of possible classifiers from which the learning algorithm chooses.
Approximation Error and Estimation Error — Boosting is a practical algorithmic paradigm for managing model complexity.
Arithmetic Operations with Numbers — input() pauses your program and waits for the user to type something on their keyboard, then returns what they typed as a string.
Arithmetic Operators — Subtraction removes the right operand from the left; unary minus (a single minus with no left operand) negates a number.
Arithmetic Operators and Expressions — Most statements (logical lines) that you write will contain expressions . A simple example of an expression is 2 + 3 . An…
Arithmetic Operators in Java — Syntax is the form a program must take, and the source stresses that knowing syntax is only part of being a successful programmer.
Arithmetic Operators in Programming — When operators have equal precedence, they are evaluated from left to right, not right to left.
Arithmetic Operators in Python — Operators are symbols or keywords that perform actions; operands are the data that operators work with.
Array Basics — An array holds a group of values, and each value sits at a numbered position.
Array Basics and Indexing — Parallel arrays store one data item across several arrays, and the same index selects that item's value in each array.
Array Basics and Syntax — An array is a numbered sequence of elements that all share one base type, and the array as a whole has type btype[].
Array Basics in Java — Arrays.copyOf(list, lengthOfCopy) returns a new array of the requested length, copying items from the original and filling any…
Array Declarations and Creation — An array literal, also called a list initializer, is a brace-enclosed list that gives initial values to a newly declared array…
Array Details and Advanced Concepts — A for-each loop is an alternative form of the for loop that makes it easier to visit every value in an array or collection.
Array Index Bounds and Exceptions — A dynamic array lets a collection grow and shrink while the program runs, and the class design hides the internal array behind…
Array Indexing and Bounds — An array's length is fixed, so a counter variable is needed to track how many positions currently hold valid items.
Array Indexing and Index Expressions — An array index can be an integer variable or an integer-valued expression, so the element accessed can change as the variable…
Array Initialization in Java — An array lets a program reach any element directly through its index, and Java indices start at 0.
Array processing with loops — Arrays.fill sets every element, or a range of elements, to one value. In the range form the starting index is included and the…
Array Variables and References — Assigning one array variable to another copies the reference, so both variables refer to the same array object.
Arrays — An interface can be the base type of an array, as in Strokeable[] listOfFigures, and its elements can refer to any objects whose…
Arrays and Index Bounds — A dynamic array in this source is a backing int array called items plus a count called itemCt, and the two are kept separate so…
Arrays and Index Positions — Unsorting is the reverse goal of sorting: it aims for a random arrangement rather than an ordered one.
Arrays and Index Positions in Java — list.size() returns the number of items in the list, which can be zero, and the valid positions are 0 through list.size()-1.
Arrays and Indexing — A variable arity method accepts a varying number of actual parameters, and the values for that parameter are placed into an…
Arrays and Memory Layout in Java — A for-each loop processes the values in an array, not the elements themselves, so assigning to the loop variable does not change…
Arrays as Objects and References — Java does not have true two-dimensional arrays. A two-dimensional array is an array whose elements are one-dimensional arrays.
Arrays as Objects in Java — An array variable holds a reference to an array object and never holds the array itself; the array object lives in the heap.
Arrays in Java — Declaring an array of objects uses the class name as the base type, such as Student[] classlist.
Arrays of Integers — A subroutine parameter lets the same subroutine perform its task on different inputs, with the caller choosing the input on each…
Arrays of Objects — Once an array of objects is filled, classlist[i] refers to a Student object, and it can be used anywhere a Student variable…
Artifacts for Capturing Per-Customer Learning — An ontology is the set of nouns and verbs that describe how an organization works, and the vendor who settles disagreements over…
Artificial Neural Networks — Backpropagation alternates forward and backward passes through an ANN with hidden layers.
Artistic Creation with AI — Generative deep learning creates new content that resembles patterns in existing data.
Asking Effective Questions About Work — A described workflow is the account of how work is supposed to happen, while an actual workflow is how it really happens.
Asking Technical Questions About the Book — The private forum is a support resource included with the purchase of Deep Learning with Python.
Assessing LLM-Generated Text Quality — Computing training and validation set losses is a way to assess the quality of LLM-generated text during training.
Assessing Vendor Capability — Slide decks and case studies describe a vendor's work but do not demonstrate a running system, and the source treats them as…
Assignment and Type Conversion in Java — A formal parameter is a named slot in a subroutine definition, with a simple identifier as its name and a declared type.
Assignment Operators — How It Works Here's how this program works. First, we assign the literal constant value 5 to the variable i using the assignment…
Assignment Statements — ++ increments a variable by 1, and -- decrements a variable by 1.
Assignment Statements and Variables — An array literal, also called a list initializer, is a brace-enclosed list that gives initial values to a newly declared array…
Associating Keyboard Accelerators with Menu Items — A keyboard accelerator is the keyboard equivalent of a menu command, and using it produces the same result as selecting the…
Associative Reinforcement Learning — The historical development of research on teams of reinforcement learning agents is commonly described in three phases.
Associative Reward-Penalty Algorithm — The historical development of research on teams of reinforcement learning agents is commonly described in three phases.
Associative Strength in Conditioning Models — The TD model treats temporal differences in prediction as learning signals.
Asymptotic Analysis of Algorithms — Θ(n²) and Θ(n³) describe growth trends, and asymptotic analysis compares algorithms by those trends for large problem sizes.
Asynchronous Events — The CPU's normal behavior is a sequential fetch-and-execute process.
Atomic Operations in Java — Threads may cache shared data, so one thread can keep seeing an old value after another thread has changed it.
Attaching ActionListeners to Components — The Action interface represents an action together with properties such as a name, an icon, and an enabled or disabled status.
Attention Dropout for Training LLMs — Attention scores above the main diagonal are masked by replacing them with negative infinity, using a triangle mask built with…
Attention Mechanism in Transformers — Building an LLM happens in three stages: implementing the architecture and preparing data, pretraining, and fine-tuning.
Attention Mechanisms in LLMs — A self-attention mechanism computes the context vector representation as a weighted sum over the inputs.
Attention Scores and Weights — The SelfAttention_v1 class stores three trainable matrices, W_query, W_key, and W_value, each mapping the input dimension to the…
Attention Weights — A context vector is the weighted sum of all input vectors, where each input is multiplied by its attention weight for one chosen…
Attention Weights and Scaled Dot-Product Attention — A causal attention mask is used when an LLM reads and generates text from left to right, to prevent access to future tokens.
Attention weights for a single input token — All context vectors are computed with one matrix multiplication: attn_weights @ inputs.
Attributes and Behaviors of Objects — Object-oriented programming designs a program around a set of objects that model the entities in a problem, rather than around a…
Attributes and Instance Variables — A method is a function that belongs to a class and includes a self parameter that refers to the specific object the method is…
Attributes and Methods — Remember, that you must refer to the variables and methods of the same object using the self only . This is called an attribute…
Audit Logging for Clinical Systems — Role-based access control restricts what a clinician can reach according to their role, and in this implementation it was…
Audit Logs for AI Agents — In regulated environments the security team examines the identity layer most closely, so an FDE needs a working answer for how…
Audit Trail Requirements — A shared service account can compromise the compliance posture for material non-public information because reads run under an…
Audit Trails in Quality Management — A high completeness score only shows that required fields are present; it does not show that the values are correct.
Authentication and API Security — An API is a published contract between applications that specifies the exact rules for requesting and receiving services.
Authentication and Authorization Basics — API keys are the mechanism vendors use to identify users and monitor their service consumption.
Authentication and Security in APIs — An API is an application-to-application contract that publishes rules for accessing services provided by one program for use by…
Authorization and Access Control — Vendors require cryptographically signed messages to verify both the source of requests and that they have not been altered in…
Autoencoders — A VAE combines an encoder, a sampling step, and a decoder.
Automatic Differentiation with Autograd — PyTorch is organized into three broad components: a tensor library, an automatic differentiation engine, and a deep learning…
Automatic Primary Keys and AUTOINCREMENT — last_insert_rowid() is a built-in SQLite function that returns the ID of the most recently inserted row in your current database…
Automating Inspection Reporting for Precision Aerospace — A high completeness score only shows that required fields are present; it does not show that the values are correct.
Automating Network Tasks with Shell Scripts — curl and wget are command-line tools for retrieving files from the web on Unix-like systems using HTTP or FTP protocols.
Autonomy Levels and Platform Access in Delta Work — Engineering momentum is one limit on how fast a solution can grow, but a customer's institutional processes are another limit…
Autonomy Levels in Agent Systems — Agent readiness is a measure of how many deterministic validation loops, such as linters, type checkers, tests, and security…
Average Reward Formulation — Differential value functions are the average-reward counterparts of familiar value functions.
Average Reward Formulation and Semi-gradient Methods — Episodic n-step semi-gradient Sarsa is based on the forward Sarsa(λ) algorithm of van Seijen (2017).
Average Reward Methods — The average reward objective evaluates average reward over a policy's state distribution.
Average-Reward Methods in Continuing Problems — Discounting does not automatically create a meaningful preference in a continuing problem with no distinguished beginning or end.
Average-Reward Reinforcement Learning — Differential values are defined with respect to the average reward.
Avoiding Global State in Functions — Python localizes variables by default on assignment, making it impossible to modify global variables from functions without the…
Axiomatic Definitions — Kleinberg's work is presented as an axiomatic attempt to define clustering.
Axis-Aligned Rectangles — The task is binary face recognition: classify an image as a human face or not.