Symmetry-based learning needs interaction with the environment.
problem Defining disentangled representations in dynamic environments.
method Building on Symmetry-Based Disentangled Representation Learning, we argue that agents need to interact with the environment to discover symmetries.
result Agents need to interact with the environment to fully understand symmetries and learn disentangled representations.
ReOPD uses pre-collected teacher trajectories to distill knowledge from multi-turn interactions.
problem The cost of fully online on-policy distillation for multi-turn interactions.
method ReOPD, an off-environment alternative that reuses pre-collected teacher trajectories as replayed prefixes, addressing the prefix trap and distribution shift.
result ReOPD preserves or improves OPD-level accuracy, uses zero tool calls, and is at least 4imes faster per training step. The paper proposes a method to learn from both simulation and real-world data.
problem Training autonomous systems in simulation and applying them to real-world environments.
method Balancing samples from simulation and real-world data using a replay buffer.
result The method achieves better performance in real-world tasks compared to training only in simulation.
KGRL uses reinforcement learning with knowledge graphs for better interactive recommendation.
problem Achieving responsiveness and accuracy in dynamic user-item interactions.
method KGRL combines reinforcement learning and knowledge graphs, using a local knowledge network and attention mechanism.
result KGRL outperforms state-of-the-art methods in simulated and real-world environments.
HRL4IN tackles interactive navigation tasks with mobile manipulators, improving efficiency and performance.
problem Interactive Navigation tasks require mobile manipulators to perform various actions, but choosing the right part of the embodiment is inefficient.
method HRL4IN uses a hierarchical reinforcement learning architecture to handle heterogeneous phases of navigation and manipulation, selecting the appropriate part of the embodiment for each phase.
result HRL4IN significantly outperforms flat PPO and HAC in terms of task performance and energy efficiency.
Study evaluates self-supervised representations in interactive environments.
problem Unclear which self-supervised methods best capture meaningful features of diverse environments.
method Quantitative evaluation of representations in two visual environments: Flappy Bird and Sonic The Hedgehog.
result Representations' utility depends on the visuals and dynamics of the environment.
OtoWorld helps agents learn to navigate by listening in interactive environments.
problem Training agents to navigate using auditory information.
method Interactive environment with agents learning to listen and navigate.
result Agents can win at OtoWorld, a navigation game with sound sources.
PHASE dataset simulates complex social interactions in physical environments.
problem Lack of datasets for evaluating physically grounded perception of complex social interactions.
method Created PHASE dataset of 2D animations with procedural generation and physics engine.
result SIMPLE model outperforms neural networks in recognizing complex social interactions.
INTAGS uses interactive simulation to improve realism in multi-agent systems.
problem Challenges in developing realistic agent-based simulators for multi-agent systems.
method INTAGS introduces a novel metric to evaluate the difference between real and synthetic multi-agent systems, optimizing a stochastic policy in reinforcement learning to adapt to interactive sequential decision-making environments.
result INTAGS generates more realistic market data compared to state-of-the-art approaches.
The rise in online social networking has brought about a revolution in social relations. However, its effects on offline interactions and its implications for collective well-being are still not clear and are under-investigated. We study the ecology of online and offline interaction in an evolutionary game framework wh…
The paper proposes a method to improve reinforcement learning by ensuring consistency between observed and imagined dynamics.
problem Compounding errors in traditional model-based reinforcement learning approaches.
method An auxiliary cost function to ensure consistency between observed and imagined dynamics.
result The proposed approach helps train powerful policies and better dynamics models.
Assisted by neural networks, reinforcement learning agents have been able to solve increasingly complex tasks over the last years. The simulation environment in which the agents interact is an essential component in any reinforcement learning problem. The environment simulates the dynamics of the agents' world and henc…
New RL method tackles sim-to-real gap using interactive data collection.
problem Sim-to-real gap in reinforcement learning.
method Distributionally robust reinforcement learning with interactive data collection.
result Proves sample-efficient learning is impossible without additional assumptions.
Complex environments and tasks pose a difficult problem for holistic end-to-end learning approaches. Decomposition of an environment into interacting controllable and non-controllable objects allows supervised learning for non-controllable objects and universal value function approximator learning for controllable obje…
The enactive approach to cognition is typically proposed as a viable alternative to traditional cognitive science. Enactive cognition displaces the explanatory focus from the internal representations of the agent to the direct sensorimotor interaction with its environment. In this paper, we investigate enactive learnin…
Interactive machine learning improves deep RL in Minecraft by giving action advice.
problem Training deep RL agents in high-aliasing environments like Minecraft is computationally expensive.
method Conducted experiments with two RL algorithms, Feedback Arbitration, and Newtonian Action Advice, to give action advice to human teachers.
result Action advice from human teachers can improve agent performance in high-aliasing environments.
VR enables professionals to develop deep learning models by moving virtual objects.
problem Challenges in understanding and developing deep learning models.
method Built a VR-based DL development environment where users interact with tangible objects to construct neural networks.
result Users can develop and understand DL models intuitively through VR, with real-time accuracy feedback.
Unified approach detects traffic conflicts across various interactions.
problem Inconsistent detection of traffic conflicts across different interactions.
method Unified probabilistic approach decomposes conflicts into statistical learning tasks.
result Effective collision warnings across diverse datasets and environments.
Adversarial methods for imitation learning have been shown to perform well on various control tasks. However, they require a large number of environment interactions for convergence. In this paper, we propose an end-to-end differentiable adversarial imitation learning algorithm in a Dyna-like framework for switching be…
Framework tackles OOD challenges in molecule property prediction by modeling environments.
problem Challenges in modeling OOD samples for molecule property prediction.
method Soft causal learning framework incorporating chemistry theories and cross-attention mechanisms.
result Demonstrates well generalization ability on seven datasets.
Neural production systems learn visual dynamics by applying rule templates to entities.
problem Modeling interactions among entities in structured visual environments.
method Inspired by production systems, the paper uses rule templates to bind placeholder variables to specific entities, scoring and applying the best fitting rules to update entity properties.
result The architecture achieves robust future-state prediction and extrapolation from simple to complex environments, outperforming GNNs.
SPARTAN learns sparse interaction graphs between objects in scenes.
problem Capturing structured interactions between entities in complex settings remains challenging.
method SPARTAN is a Transformer-based world model that learns sparse, context-dependent interaction graphs through sparsity regularization on attention patterns.
result SPARTAN achieves significantly improved few-shot adaptation to dynamics changes and robustness against distractors.
RecSim creates customizable simulation environments for RSs.
problem Creating realistic simulation environments for RSs.
method Configurable platform for authoring simulation environments.
result Enables pushing the limits of RL and RS techniques.
Autonomous robots need to interact with unknown, unstructured and changing environments, constantly facing novel challenges. Therefore, continuous online adaptation for lifelong-learning and the need of sample-efficient mechanisms to adapt to changes in the environment, the constraints, the tasks, or the robot itself a…
Modular method predicts motion in crowded scenes using learned environment models.
problem Predicting motion in dynamic, crowded environments.
method Modular model of spatial and dynamic aspects, unsupervised adaptation to new tasks.
result Comparable performance to state-of-the-art, transferable across tasks.
Behavioral cloning fails due to ignoring causal structure, leading to worse performance.
problem Behavioral cloning's failure to account for causal structure causes worse performance.
method Investigates and proposes interventions to correct causal misidentification.
result Causal misidentification occurs in various domains and can be mitigated.
Investigates safe decision-making in interactive environments.
problem Learning the best safe decision in real-time systems.
method Reduces to a constrained linear bandits problem, proposes adaptive experimental design-based algorithm.
result First results on best-arm identification in linear bandits with safety constraints.
Interactive machine comprehension models learn through seeking relevant information.
problem Existing MRC models struggle with real-world applications due to static datasets.
method Reframe MRC datasets as interactive, partially observable environments.
result Interactive models can seek relevant information through sequential decision making.
A generative recurrent neural network is quickly trained in an unsupervised manner to model popular reinforcement learning environments through compressed spatio-temporal representations. The world model's extracted features are fed into compact and simple policies trained by evolution, achieving state of the art resul…
Dreamer 4 learns Minecraft tasks from videos alone.
problem Accurately predicting object interactions in complex environments.
method Reinforcement learning inside a fast, accurate world model.
result Dreamer 4 outperforms previous models in Minecraft, learning from only offline data.
In this paper we study a new reinforcement learning setting where the environment is non-rewarding, contains several possibly related objects of various controllability, and where an apt agent Bob acts independently, with non-observable intentions. We argue that this setting defines a realistic scenario and we present …
Fine-tuning RL with offline data reduces online interactions.
problem Optimizing RL with limited online interactions and offline data.
method Developed algorithm extsc{FTPedel} for MDPs with linear structure.
result Optimally reduces the number of online interactions needed.
New snooping attacks exploit deep RL without access to environment.
problem Security vulnerabilities in deep reinforcement learning.
method Proposes snooping threat models and attacks on RL agents.
result Adversaries can launch attacks without interacting with the environment.
PPOPT uses pretraining to speed up reinforcement learning in physics simulations.
problem High computational costs and inefficiency in reinforcement learning with small training samples.
method A novel policy neural network architecture that combines pretraining and fully-connected networks.
result PPOPT outperforms classic PPO on small training samples in terms of rewards and stability.
Recent advances in artificial intelligence have been driven by the presence of increasingly realistic and complex simulated environments. However, many of the existing environments provide either unrealistic visuals, inaccurate physics, low task complexity, restricted agent perspective, or a limited capacity for intera…
Two algorithms improve Federated RL in diverse environments.
problem Collaborative learning in environments with varying dynamics.
method Proposed two federated RL algorithms, QAvg and PAvg, and a personalization heuristic.
result Achieved better performance and generalization in diverse environments.
Proposes learning task-agnostic dynamics priors for faster RL.
problem Challenges in learning accurate dynamics models for RL.
method Pre-training a frame predictor on physics videos to initialize and fine-tune dynamics models.
result Improves policy learning and convergence, outperforming competitors.
For sophisticated reinforcement learning (RL) systems to interact usefully with real-world environments, we need to communicate complex goals to these systems. In this work, we explore goals defined in terms of (non-expert) human preferences between pairs of trajectory segments. We show that this approach can effective…
It has been postulated that a good representation is one that disentangles the underlying explanatory factors of variation. However, it remains an open question what kind of training framework could potentially achieve that. Whereas most previous work focuses on the static setting (e.g., with images), we postulate that…
EduGym simplifies reinforcement learning education for students.
problem Students struggle to translate between equations and code in reinforcement learning.
method EduGym provides educational environments and interactive notebooks.
result 86% of students and researchers find EduGym useful for education.
Self-Predictive Representations improves data-efficient reinforcement learning from limited interaction.
problem Efficient reinforcement learning from limited data.
method Train agents to predict future latent state representations using self-supervised objectives.
result Achieves a median human-normalized score of 0.415 on Atari with 100k steps of interaction, 55% improvement over previous state-of-the-art.
Improved model-based reinforcement learning for interactive dialogue tasks reduces sample needs and improves performance.
problem Limited data and high sample cost in interactive dialogue systems.
method Model-based actor-critic approach with an environment model and planner.
result 70 times fewer samples required compared to baseline model-free algorithm, with 2x better asymptotic performance.
We identify two issues with the family of algorithms based on the Adversarial Imitation Learning framework. The first problem is implicit bias present in the reward functions used in these algorithms. While these biases might work well for some environments, they can also lead to sub-optimal behavior in others. Secondl…
We propose a method for modeling and learning turn-taking behaviors for accessing a shared resource. We model the individual behavior for each agent in an interaction and then use a multi-agent fusion model to generate a summary over the expected actions of the group to render the model independent of the number of age…
Paper proposes a deep RL method for hedging variable annuities, outperforming misspecified models.
problem Model miscalibration in variable annuity contracts with GMMB and GMDB riders.
method Two-phase deep reinforcement learning approach: training phase in a controlled environment, online learning phase in real market.
result Trained reinforcement learning agent hedges equally well as correct Delta in training phase and outperforms misspecified Deltas.
MIDAS learns to adaptively control other cars in urban driving scenarios.
problem Autonomous vehicles need to interact with other agents on the road.
method Reinforcement learning with attention mechanism to handle multiple agents.
result MIDAS policies are adaptive and robust to external changes.
WayDCM predicts trajectories considering long-term goals, improving accuracy.
problem Predicting future trajectories of dynamic agents in complex environments.
method WayDCM combines DCM and NN to predict intermediate goals and trajectories, considering long-term goals.
result WayDCM outperforms previous methods on the Waymo Open dataset.
Infants are experts at playing, with an amazing ability to generate novel structured behaviors in unstructured environments that lack clear extrinsic reward signals. We seek to mathematically formalize these abilities using a neural network that implements curiosity-driven intrinsic motivation. Using a simple but ecolo…