Stop Regressing: Training Value Functions via Classification for Scalable Deep RL.
Jesse FarebrotherJordi OrbayQuan VuongAdrien Ali TaïgaYevgen ChebotarTed XiaoAlex IrpanSergey LevinePablo Samuel CastroAleksandra FaustAviral KumarRishabh AgarwalPublished in: CoRR (2024)
Keyphrases
- training phase
- training set
- training samples
- supervised learning
- reinforcement learning
- training process
- support vector
- decision trees
- pattern classification
- classification method
- classification algorithm
- classification performances
- classification accuracy
- classification models
- unsupervised learning
- classification scheme
- pattern recognition
- feature vectors
- training examples
- preprocessing
- multi agent
- classification rules
- multi layer perceptron
- deep architectures
- machine learning
- reinforcement learning algorithms
- training patterns
- discriminative training
- automatic classification
- decision rules
- test set
- optimal policy
- machine learning algorithms
- text classification
- support vector machine
- hidden markov models
- training data