Methods

Algorithms and Methods

@@Home -> ArtificialIntelligenceDictionary -> methods

---

General Methods

• back-propagation (Werbos, 1974)
• backward modeling
• batch training

  *# batch back-propagation
  *# training by epoch

• boosting (Schapire, 2001)
• cascade algorithm
• constructive training (of neural network)
• convex optimization
• covariance training
• direct error minimization
• discriminative training methods
• dynamic algorithms
• ensemble learning (Krogh and Vedelsby, 1995; Diettrich, 2000)
• explorative algorithms
• exploitive algorithms
• forward modeling
• genetic algorithms (Goldberg, 1989)
• global optimization techniques
• gradient methods
• greedy learning
• incremental training (of neural network)

  *# incremental back-propagation
  *# online training
  *# training by pattern

• kernel methods
• learning

  *# learning-by-example

• mini-batch training (Wilson and Martinez, 2003)
• off-line learning
• off-policy learning
• on-policy learning
• online learning
• policy iteration algorithm (Kaelbling et al., 1996)
• reinforcement learning

  *# episodic reinforcement learning
  *# model-based reinforcement learning
  *# model-free reinforcement learning
  *# non-episodic reinforcement learning
  *# off-policy learning
  *# on-policy learning
  *# tabular reinforcement learning

• supervised training
• temporal difference learning algorithms
• training
• unsupervised training
• value iteration algorithm
• wake-sleep algorithm

  *# contrastive wake-sleep

• weight update algorithm

Common Parameters of Algorithms

• fixed step-size
• dynamic step-size
• patience parameter
• momentum
• steepness parameter
• step size

  *# fixed step-size
  *# dynamic step-size

• variational bound

Named Algorithms

• Backprop
• Bayesian techniques (Neal, 1996)
• Cascade 2

  *# Cascade 2 with caching

• Cascade Correlation (Prechelt, 1997)
• Casper algorithm (Treadgold and Gedeon, 1997)
• Cerebellar Model Articulation Controller (CMAC) (Albus, 1975; Glanz et al., 1991; Sutton and Barto, 1998)
• Contrastive Divergence Learning
• Dyna-Q (Sutton and Barto, 1998)
• Explicit Explore or Exploit (Kearns and Singh, 1998)
• Gibbs Sampling

  *# Alternating Gibbs Sampling

• K-Nearest Neighbor
• Learning Vector Quantization (LVQ)
• Levenberg-Marquardt (More, 1977)
• Locality-Sensitive Hashing (LSH)
• Maximum Likelihood (ML) Learning
• Model-Based Interval Estimation (MBIE) (Strehl and Littman, 2004)
• Model-Based Policy Gradient methods (MBPG) (Wang and Dietterich, 2003)
• Monte Carlo Algorithms
• N-Step Return Algorithm
• Neural Fitted Q Iteration (Riedmiller, 2005)

  *# NFQ-SARSA(L)

• Optimal Brain Damage (LeCun et al., 1990)
• Orthogonal Least Squares (OLS)
• Particle Swarm (Kennedy and Eberhart, 1995)
• Prioritized Sweeping (Sutton and Barto, 1998)
• Q-Learning

  *# Delayed Q-learning (Strehl et al., 2006)
  *# Generalized Policy Iteration (GPI) (Sutton and Barto, 1998)
  *# Naive Q(λ) (Sutton and Barto, 1998)
  *# One-Step Q-Learning
  *# Peng’s Q(λ) (Peng and Williams, 1994)
  *# Q(λ)
  *# Watkin’s Q(λ) (Watkins, 1989)

• Q-SARSA(λ)

  *# One-Step Q-SARSA Algorithm

• Quickprop (Fahlman, 1988)
• R-Max (Brafman and Tennenholtz, 2002)
• RPROP (Riedmiller and Braun, 1993)

  *# iRPROP- (Igel and Hsken, 2000)

• SARSA(λ)

  *# On-Policy SARSA-Learning
  *# SARSA with Linear Function Approximators (Gordon, 2000)

• Simulated Annealing (Kirkpatrick et al., 1987)
• Sliding Window Cache
• SONN (Tenorio and Lee, 1989)
• Temporal Difference (TD) Learning

  *# Monte Carlo Prediction
  *# TD(0) Algorithm (Sutton, 1988)
  *# TD(λ) Algorithm
  *# TD(λ) Algorithm With Linear Function Approximators (Tsitsiklis and Roy, 1996)
  *# Temporal Difference (TD) Prediction Algorithm

• λ-Return Approach