#reinforcement learning
25 resultsRLTM: An Efficient Neural IR Framework for Long Documents
Chen Zheng, Yu Sun, Shengxian Wan +1
The paper introduces RLTM, a neural ranking framework that first selects relevant sentences from long documents with a fast model and then applies a more sophisticated model to com…
Learning to Explore in Motion and Interaction Tasks
Miroslav Bogdanovic, Ludovic Righetti
The paper proposes a generative exploration strategy that reuses data from previously solved robotic tasks to speed up model-free reinforcement learning for new motion and contact…
Trajectory-wise Control Variates for Variance Reduction in Policy Gradient Methods
Ching-An Cheng, Xinyan Yan, Byron Boots
The paper proposes trajectory-wise control variates to more effectively reduce the variance of Monte Carlo gradient estimates in policy gradient reinforcement learning, showing the…
Cost-Aware Fine-Grained Recognition for IoTs Based on Sequential Fixations
Hanxiao Wang, Venkatesh Saligrama, Stan Sclaroff +1
The paper introduces DRIFT, a deep reinforcement learning model that performs sequential, foveated fixations on images to enable fine‑grained classification on low‑power edge camer…
Goal-Driven Sequential Data Abstraction
Umar Riaz Muhammad, Yongxin Yang, Timothy M. Hospedales +2
The paper proposes a reinforcement‑learning framework that learns to create compact abstractions of sequential data (e.g., sketches, videos, text) in a goal‑driven manner without n…
Incremental Reinforcement Learning --- a New Continuous Reinforcement Learning Frame Based on Stochastic Differential Equation methods
Tianhao Chen, Limei Cheng, Yang Liu +2
The paper introduces Incremental Reinforcement Learning (IRL), a continuous reinforcement‑learning framework built on stochastic differential equations that ensures action continui…
Improving Skin Condition Classification with a Visual Symptom Checker Trained using Reinforcement Learning
Mohamed Akrout, Amir-massoud Farahmand, Tory Jarmain +1
The paper introduces a visual symptom checker that combines a pre‑trained CNN with a reinforcement‑learning agent to ask patients targeted questions, improving skin condition class…
Task-Oriented Optimal Sequencing of Visualization Charts
Danqing Shi, Yang Shi, Xinyue Xu +4
The paper proposes a reinforcement‑learning method to automatically generate optimal sequences of visualization charts that support specific analysis tasks such as correlation, ano…
Developing a Simple Model for Sand-Tool Interaction and Autonomously Shaping Sand
Wooshik Kim, Catherine Pavlov, Aaron M. Johnson
The paper presents a simplified model of how a robot tool interacts with sand using a surface heightmap, and demonstrates methods to autonomously shape sand into desired forms via…
Adaptive Stress Testing with Reward Augmentation for Autonomous Vehicle Validation
Anthony Corso, Peter Du, Katherine Driggs-Campbell +1
The paper enhances Adaptive Stress Testing for autonomous vehicle validation by adding reward augmentation to guide reinforcement‑learning searches toward a broader set of realisti…
Adaptive Critic Based Optimal Kinematic Control for a Robot Manipulator
Aiswarya Menon, Ravi Prakash, Laxmidhar Behera
The paper presents a new weight‑update law for a single‑network adaptive critic that guarantees convergence to the optimal cost and stability in kinematic control of robot manipula…
Attention Control with Metric Learning Alignment for Image Set-based Recognition
Xiaofeng Liu, Zhenhua Guo, Jane You +1
The paper proposes a reinforcement‑learning based attention control network that models dependencies among unordered images in a set for face verification and identification, and i…
Speech Driven Backchannel Generation using Deep Q-Network for Enhancing Engagement in Human-Robot Interaction
Nusrah Hussain, Engin Erzin, T. Metin Sezgin +1
The paper introduces a batch reinforcement learning approach using a deep Q-network to train a social robot to produce non‑verbal backchannels, such as laughs, that increase user e…
Ego-Pose Estimation and Forecasting as Real-Time PD Control
Ye Yuan, Kris Kitani
The paper introduces a reinforcement‑learning based proportional‑derivative control policy that estimates and forecasts 3D human pose from egocentric video in real time.
Flying through a narrow gap using neural network: an end-to-end planning and control approach
Jiarong Lin, Luqi Wang, Fei Gao +2
The paper introduces an end‑to‑end neural network that directly maps a drone’s perception of a tilted narrow gap to thrust and attitude commands, using imitation learning followed…
Time-Optimal Path Tracking for Industrial Robots: A Dynamic Model-Free Reinforcement Learning Approach
Jiadong Xiao, Lin Li, Tie Zhang +1
The paper proposes a model‑free reinforcement‑learning method (TOPTO‑SARSA) to compute time‑optimal trajectories for industrial robot manipulators while respecting kinematic constr…
CARL: Aggregated Search with Context-Aware Module Embedding Learning
Xinting Huang, Jianzhong Qi, Yu Sun +2
The paper proposes CARL, a model that uses attention over top-ranked blue-links (the context) to learn context-aware embeddings for heterogeneous search modules, optimizing module…
Variable Impedance Control in End-Effector Space: An Action Space for Reinforcement Learning in Contact-Rich Tasks
Roberto MartÃn-MartÃn, Michelle A. Lee, Rachel Gardner +3
The paper introduces Variable Impedance Control in End-effector Space (VICES) as an action space for deep reinforcement learning in contact-rich manipulation tasks, demonstrating i…
Large scale continuous-time mean-variance portfolio allocation via reinforcement learning
Haoran Wang
The paper introduces a reinforcement‑learning approach for large‑scale continuous‑time mean‑variance portfolio allocation, using an exploratory control framework and a Gaussian fee…
Robby is Not a Robber (anymore): On the Use of Institutions for Learning Normative Behavior
Stevan Tomic, Federico Pecora, Alessandro Saffiotti
The paper introduces a framework that encodes human social norms as rewards to guide reinforcement learning agents toward normative behavior, allowing the same norms to be applied…
Learning Real-World Robot Policies by Dreaming
AJ Piergiovanni, Alan Wu, Michael S. Ryoo
The paper proposes a learned, action‑conditioned world model that generates realistic image sequences (“dreams”) so a robot can train visuomotor policies in simulation and then tra…
Learning When to Drive in Intersections by Combining Reinforcement Learning and Model Predictive Control
Tommy Tram, Ivo Batkovic, Mohammad Ali +1
The paper presents a two‑level decision system for autonomous vehicles at intersections, combining a high‑level reinforcement‑learning policy with a low‑level model predictive cont…
Taxable Stock Trading with Deep Reinforcement Learning
Shan Huang
The paper uses deep reinforcement learning to develop stock trading strategies that account for capital gains taxes and tax rebates, showing that ignoring taxes can reduce portfoli…
Reinforcement Learning with Wasserstein Distance Regularisation, with Applications to Multipolicy Learning
Mohammed Amin Abdullah, Aldo Pacchiano, Moez Draief
The paper introduces a regularizer based on the Wasserstein distance between distributions of trajectory embeddings to train reinforcement‑learning policies that are either distinc…