Proposes learning latent reward model for planning from rewards.
problem Planning in high-dimensional state spaces with limited reward information.
method Directly learns a latent dynamics model from rewards, planning in latent state-space.
result Successfully learns accurate latent reward prediction model, achieving strong performance and high sample efficiency.
Proposes DLGPD model to learn dynamics from images for planning.
problem Planning in unknown, indirectly observable environments.
method Deep latent Gaussian process dynamics model trained jointly with neural networks.
result Demonstrates improved data efficiency and transfer learning.
Planning has been very successful for control tasks with known environment dynamics. To leverage planning in unknown environments, the agent needs to learn the dynamics from interactions with the world. However, learning dynamics models that are accurate enough for planning has been a long-standing challenge, especiall…
Regularized OT improves disentangled latent representations in GANs.
problem Disentangled representation learning in GANs.
method Structured regularization of optimal transport for latent space.
result Regularization leads to informative latent dimensions in GANs.
XLVINs improve data efficiency in implicit planning by leveraging latent space.
problem Improving data efficiency in implicit planning algorithms.
method XLVINs use a high-dimensional latent space to perform planning computations, breaking the algorithmic bottleneck.
result XLVINs significantly improve data efficiency across various settings compared to value iteration-based implicit planners and model-free baselines.
We demonstrate the use of conditional autoregressive generative models (van den Oord et al., 2016a) over a discrete latent space (van den Oord et al., 2017b) for forward planning with MCTS. In order to test this method, we introduce a new environment featuring varying difficulty levels, along with moving goals and obst…
Certified guidance ensures generative models always meet planning objectives.
problem Lack of guarantees for generated outputs satisfying planning objectives.
method Certified guidance modifies generative models without retraining, ensuring outputs meet STL specifications.
result Certified guidance produces models that are always correct, unlike non-certified methods.
Director learns hierarchical behaviors from pixels, outperforming exploration methods.
problem Challenges in learning long-term hierarchical behaviors from pixels.
method Directly learns hierarchical behaviors from pixels using a latent space world model.
result Director outperforms exploration methods on various tasks.
Enhances PlaNet for better planning in uncertain environments.
problem Improving deep planning networks for partially observable environments.
method Incorporates Bayesian inference to handle uncertainty in latent models and action candidates.
result Consistently improves asymptotic performance on continuous control tasks.
In model-based reinforcement learning, the agent interleaves between model learning and planning. These two components are inextricably intertwined. If the model is not able to provide sensible long-term prediction, the executed planner would exploit model flaws, which can yield catastrophic failures. This paper focuse…
A method uses RL to learn abstractions for planning, improving robot navigation and manipulation tasks.
problem Planning requires suitable abstractions for states and transitions, which RL struggles with for temporally extended tasks.
method Goal-conditioned policies learned with RL are incorporated into planning, with a latent variable model representing valid states.
result Our method significantly outperforms prior work on image-based robot navigation and manipulation tasks.
Framework simplifies vision-based control and goal discovery.
problem Learning proportional control from visual data.
method Introduces NewtonianVAE for proportional control and goal discovery.
result Dramatic simplification and acceleration of vision-based controllers.
A key challenge in complex visuomotor control is learning abstract representations that are effective for specifying goals, planning, and generalization. To this end, we introduce universal planning networks (UPN). UPNs embed differentiable planning within a goal-directed policy. This planning computation unrolls a for…
LeJEPA learns latent variables from nonlinear observations.
problem Learning latent variables from nonlinear observations.
method Proves linear identifiability of Gaussian latent distributions.
result Gaussian distribution uniquely guarantees linear identifiability.
In this work, we take a representation learning perspective on hierarchical reinforcement learning, where the problem of learning lower layers in a hierarchy is transformed into the problem of learning trajectory-level generative models. We show that we can learn continuous latent representations of trajectories, which…
This work tackles long-term visual planning by goal-conditioned hierarchical predictors.
problem Current learning approaches fail on long-horizon tasks due to lack of goal information and coarse-to-fine planning.
method Formulate goal-conditioned predictors (GCPs) and hierarchical models to predict trajectories between observations.
result GCPs enable effective long-term planning with much longer horizons than before.
This paper studies when particle filtering is efficient for planning in partially observed systems.
problem The efficiency of particle filtering for planning in partially observed linear dynamical systems.
method Coupling of ideal and approximate sequences to bound particle complexity.
result Polynomially many particles suffice for stable systems to approximate optimal planning.
Plan2Vec learns image representations without labels, improving control tasks.
problem Learning image representations without labeled data.
method Constructs a weighted graph using near-neighbor distances and extrapolates to global embedding.
result Plan2Vec achieves accurate long-term value estimates in control tasks with reduced computational and memory costs.
New framework handles dynamic contexts in reinforcement learning.
problem Learning in environments where contexts change over time.
method Dynamic Contextual Markov Decision Processes (DCMDPs) with logistic aggregation.
result Upper-confidence-bound style algorithm with regret bounds.
Latent variable models improve RL by facilitating efficient learning and exploration.
problem Improving sample efficiency in reinforcement learning.
method Representation view of latent variable models for state-action value functions, incorporating kernel embeddings and UCB exploration.
result Established sample complexity of the proposed approach in online and offline settings, demonstrated superior performance in benchmarks.
Unified Latent Dynamics unifies model-free and model-based reinforcement learning.
problem Combining the efficiency of model-free methods with the representational strengths of model-based approaches.
method Embedding state-action pairs into a latent space where the true value function is approximately linear, using synchronized updates of encoder, value, and policy networks.
result ULD achieves cross-domain competence with minimal tuning and a fraction of the parameter footprint.
Integrates skills and world models for efficient task solving and transfer.
problem Quickly solve new tasks in complex environments using reusable knowledge.
method Leverages partial amortization for fast adaptation and online skill planning.
result Improved sample efficiency in single tasks and transfer between tasks.
We present a representation learning algorithm that learns a low-dimensional latent dynamical system from high-dimensional \textit{sequential} raw data, e.g., video. The framework builds upon recent advances in amortized inference methods that use both an inference network and a refinement procedure to output samples f…
Cultural activity is an inherent aspect of urban life and the success of a modern city is largely determined by its capacity to offer generous cultural entertainment to its citizens. To this end, the optimal allocation of cultural establishments and related resources across urban regions becomes of vital importance, as…
Conventionally, model-based reinforcement learning (MBRL) aims to learn a global model for the dynamics of the environment. A good model can potentially enable planning algorithms to generate a large variety of behaviors and solve diverse tasks. However, learning an accurate model for complex dynamical systems is diffi…
This work uses action equivariance to learn structured latent spaces for reinforcement learning.
problem Learning structured latent spaces for reinforcement learning.
method Introduced a contrastive loss function to enforce action equivariance on learned representations.
result Optimal policies in the abstract MDP can be successfully lifted to the original MDP.
XLVINs improve deep reinforcement learning by combining self-supervised learning and neural algorithmic reasoning.
problem Limitations of Value Iteration Networks (VINs) in deep reinforcement learning.
method Combining contrastive self-supervised learning, graph representation learning, and neural algorithmic reasoning.
result XLVINs match VIN-like models on discrete, fixed MDPs and significantly outperform model-free baselines.
MIRO learns robust latent spaces by maximizing mutual information with future information.
problem Robust perception in complex, unstructured environments with low sample complexity.
method MIRO maximizes mutual information in a latent space for model-based reinforcement learning.
result MIRO outperforms reconstruction objectives in cluttered scenes.
The paper proposes a model to forecast traffic motion from sensor data.
problem Accurately predicting traffic motion for safe vehicle maneuvers.
method Implicit latent variable model using interaction graphs and graph neural networks.
result Achieves state-of-the-art motion forecasting and interaction understanding.
TASID learns policies in high-dimensional settings with abstract simulator knowledge.
problem RL in high-dimensional settings with limited observation knowledge.
method TASID algorithm for transfer RL from abstract simulator with bounded perturbations.
result Sample complexity polynomial in horizon, independent of number of states.
MuZero visualizes its internal representations to stabilize planning.
problem Stability issues in MuZero's planning process.
method Visualized MuZero's latent representations and proposed regularization techniques.
result Action trajectories diverge between observation embeddings and internal state transitions, leading to instability.
New algorithm for planning in observable POMDPs in quasi-polynomial time.
problem Planning in POMDPs is computationally hard.
method Assumption of well-separated distributions on states and observations leads to quasi-succinct descriptions of near-optimal policies.
result Quasipolynomial-time algorithm for planning in observable POMDPs.
In recent years, Generative Adversarial Networks (GAN) have emerged as a powerful method for learning the mapping from noisy latent spaces to realistic data samples in high-dimensional space. So far, the development and application of GANs have been predominantly focused on spatial data such as images. In this project,…
Intelligent agents can learn to represent the action spaces of other agents simply by observing them act. Such representations help agents quickly learn to predict the effects of their own actions on the environment and to plan complex action sequences. In this work, we address the problem of learning an agent's action…
Proves sufficiency of countable test plans for BV functions on metric spaces.
problem Recovering BV functions and their measures on arbitrary metric spaces.
method Proves sufficiency of countable test plans on arbitrary metric measure spaces and geodesics on CD(K,N) spaces. result Countable test plans are sufficient for BV functions and their measures on metric spaces.
GeMA learns latent manifolds to benchmark complex systems.
problem Benchmarking complex systems like rail networks and economies with classical methods.
method Geometric Manifold Analysis (GeMA) using a productivity-manifold variational autoencoder (ProMan-VAE).
result GeMA provides more nuanced efficiency evaluations in complex systems.
New approach handles stochastic and partially-observable environments using discrete autoencoders and Monte Carlo tree search.
problem Challenges in planning for stochastic and partially-observable environments.
method Uses discrete autoencoders and a stochastic variant of Monte Carlo tree search.
result Significantly outperforms MuZero on stochastic chess and scales to DeepMind Lab.
SPLICE generates accurate time-series imputations with reliable prediction intervals.
problem Lack of reliability guarantees in time-series imputation models.
method Modular framework combining latent generative imputation with distribution-free prediction intervals.
result SPLICE achieves lowest mean Load-only MSE and best CRPS on various datasets.
In the quest for efficient and robust reinforcement learning methods, both model-free and model-based approaches offer advantages. In this paper we propose a new way of explicitly bridging both approaches via a shared low-dimensional learned encoding of the environment, meant to capture summarizing abstractions. We sho…
Geodesics in R^n configuration spaces for points apart by epsilon.
problem Finding paths between points in R^n with minimum distance constraints.
method Explicit formulas for geodesics in configuration spaces with Euclidean metric.
result Geodesic motion-planning rules for configuration spaces of ordered pairs of points.
New method combines heuristics and search techniques to speed up cooperative planning for autonomous vehicles.
problem Efficient cooperative planning for autonomous vehicles in complex traffic scenarios.
method Combining learned heuristics with Monte Carlo Tree Search (MCTS) to guide search towards promising actions.
result Better solutions at lower computational costs achieved through accelerated planning.
A planning approach learns skills from interactions, balancing exploration and exploitation.
problem Learning robust high-level skills in noisy environments with unknown pre-conditions.
method Formulates skills as high-level policies, learns plans via bandit problems, balances exploration and exploitation.
result A planner capable of learning robust high-level skills in high-dimensional state spaces.
Predictive state representations (PSRs) offer an expressive framework for modelling partially observable systems. By compactly representing systems as functions of observable quantities, the PSR learning approach avoids using local-minima prone expectation-maximization and instead employs a globally optimal moment-base…
In this paper we study a notion of topological complexity for the motion planning problem. The topological complexity is a number which measures discontinuity of the process of motion planning in the configuration space X. More precisely, it is the minimal number k such that there are k different motion planning rules,…
Generative model for high-dimensional categorical data using Gaussian-Dirichlet fields.
problem Efficiently modeling and predicting high-dimensional categorical data.
method Combines Dirichlet and Gaussian processes for spatio-temporal modeling.
result Model accurately approximates categorical data in unobserved locations.
A method for learning a context latent vector to improve generalization in model-based RL.
problem Learning a global dynamics model that can generalize across different dynamics.
method Decomposes learning a global dynamics model into two stages: learning a context latent vector and predicting next states.
result Achieves superior generalization across various simulated robotics and control tasks.
PackIt creates a virtual space for testing geometric planning skills.
problem Evaluating geometric planning abilities in virtual environments.
method Developed a virtual environment, PackIt, for geometric planning tasks.
result Demonstrated the effectiveness of various methods for geometric planning.
This work improves AI's ability to solve physical tasks by optimizing world models in abstracted spaces.
problem Developing AI agents capable of solving diverse physical tasks and generalizing to new environments.
method Investigates and optimizes a family of joint-embedding predictive world models (JEPA-WMs) for efficient planning in abstracted spaces.
result Proposes a model that outperforms two established baselines in both navigation and manipulation tasks.