Optimization of Reinforcement Learning in Personalized Teaching Mode of College English Classroom under the OBE Concept

In current college English teaching, traditional teaching models often ignore students' personalized needs and learning differences, resulting in uneven learning results and difficulty in meeting the student-centered educational goals under the OBE (Outcome-Based Education) concept. This study adopts an experimental design. First, students' learning data, including vocabulary, grammar mastery, reading comprehension and listening level, are collected through preliminary tests. Then, a personalized learning model is constructed using the DQN algorithm. The specific steps include: initializing the Q-value table and setting the learning rate and discount factor; selecting the optimal teaching strategy based on the student's current state; executing the strategy and observing the changes in the student's learning effect; updating the Q-value table and iteratively optimizing the strategy; converging the model through multiple rounds of training. Finally, through comparative experiments, the personalized teaching model optimized by reinforcement learning (RL) significantly improves students' academic performance. The average score of the proposed method increases by 12.72 points, the average score in the vocabulary test increases by 15.12 points, the reading comprehension ability increases by 5.72 points, and the listening level increases by 5.68 points. Reinforcement learning can effectively optimize the personalized teaching model in college English classrooms under the OBE concept and significantly improve students' vocabulary, reading comprehension and listening level.