LLM-Powered Code Vulnerability Repair with Reinforcement Learning and Semantic Reward.
Nafis Tanveer IslamJoseph KhouryAndrew SeongGonzalo De La Torre ParraElias Bou-HarbPeyman NajafiradPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- eligibility traces
- function approximation
- semantic information
- state space
- high level
- model free
- markov decision processes
- machine learning
- temporal difference
- semantic web
- source code
- natural language
- domain specific
- average reward
- reinforcement learning algorithms
- learning algorithm
- semantic annotation
- optimal policy
- semantic similarity
- multi agent
- optimal control
- transfer learning
- reward function
- semantic search
- dynamic programming
- partially observable
- learning agent
- metadata
- semantic description
- damage assessment
- neural network
- partially observable environments