Adaptive stock trading strategies with deep reinforcement learning methods Wu, Xing, et al. "Adaptive stock trading strategies with deep reinforcement learning methods." Information Sciences 538 (2020): 142-158. Highlights - Gated Recurrent Unit is proposed to extract informative features from raw financial data. - Reward function is designed with risk-adjusted ratio for trading strategies for s..

Generative Question Refinement with Deep Reinforcement Liu, Ye, et al. "Generative question refinement with deep reinforcement learning in retrieval-based QA system." Proceedings of the 28th ACM International Conference on Information and Knowledge Management. 2019. Abstract 실제 QA 시스템에서 잘못된 단어, 잘못된 단어 순서, 노이즈 같은 잘못된 형식의 질문들이 일반적이여서 QA 시스템이 이를 정확하게 이해하고 답변을 던지지 못하게 만듬. 이러한 잘못된 형식의 질문을 효과적으로 제거하기 ..

Multi-DQN An ensemble of Deep Q-learning agents for stock market forecasting Carta, Salvatore, et al. "Multi-DQN: An ensemble of Deep Q-learning agents for stock market forecasting." Expert systems with applications 164 (2021): 113820. 주요 하이라이트 A novel ensembling methodology of RL agents with different training experiences. Validation of such ensemble in intraday stock market trading. Different ..

Introduction - Classification, Prediction using RL like DQN. - Different Existing deep learning is a method of increasing the accuracy of a model through a network. DQN is a reinforcement learning method in which Q values are selected and acted through a model. - Usually, RL does not used to solve classification or Prediction problems. DQN code used in the game is analyzed and refactored to be u..

DQN (Deep Q Network) DQN is learning using deep learning neural networks such as CNN. Method of storing samples obtained from each time step, randomly selecting these samples to configure and update them into mini-batches. Existing RL, once it start learning in a bad way, it continue learning in a bad way. The problem is solved by randomly extracting and breaking the correlation between samples...

MARL(Multi-Agent Reinforcement Learning) - Trying to study Multi-Agent Algorithms in reinforcement learning. - For collaboration or competition, it is a field in which multiple agents interact with each other and find optimal behavior. - In reality, in order for Reinforcement learning to actually apply, the characteristics of various fields must be considered, so multi-agent consideration is ess..

import gym import time env = gym.make('CartPole-v1') #강화학습 환경 불러오기 for i_episode in range(20): # 새로운 에피소드(initial environment)를 불러온다(reset) observation = env.reset() for t in range(100): env.render() #화면에 출력 / 행동 취하기 이전 환경에서 얻은 관찰값(obsevation)적용해서 그림 time.sleep(0.05) # 행동(action)을 취하기 이전에 환경에 대해 얻은 관찰값(observation) print('observation before action:') print(observation) action = env.action_space...

아무것도 안알려주고 일단 시도해보고 시행착오 겪으면서 실력을 키워나가게 하는 방법임. Agent가 올바른 행동을 하면 보상(rewards)을 주고, 불리한 행동을 하면 벌점을 부여해줌. 이렇게 행동 하나하나가 쌓여서 보상이 최대화가 되게 만드는 학습방법임. Agent가 자신이 잘하고 있는 것인지, 잘 못하고 있는 것인지 확실하게 알아야하기에 무조건 scalar feedback을 해야함. ex) +1, -3, +2.6, +2 ... 환경에 대한 사전지식이 없는 상태로 학습이 시작되고, 보상을 통하여 학습을 함. 어떠한 행동을 했을 때 환경이 어떻게 반응하는지에 따른 보상이 주어짐. 어떻게 보상이 최대화가 될 수 있는지 학습을 하는 것이 강화학습. Agent : 주어진 문제 내에서 행동을 하는 주체. Sta..