Neural architectures learn belief representations for partially observable environments.
problem Learning belief states in partially observable domains.
method One-step frame prediction and contrastive predictive coding (CPC) as objective functions.
result Neural architectures can learn belief representations, encoding both state information and uncertainty.
RPSP networks combine PSRs and RNNs for reinforcement learning in POE.
problem Learning in partially observable environments.
method Recurrent filter with PSR, reactive policy, gradient descent.
result RPSP networks outperform memory-preserving models.
PSDs improve RNN performance by predicting future observations.
problem Modeling dynamic processes with unknown latent states.
method Augmenting RNNs with Predictive-State Decoders (PSDs) that target predicting future observations.
result PSDs improve statistical performance of state-of-the-art RNNs with fewer iterations and less data.
Predictive State Representations (PSRs) are an expressive class of models for controlled stochastic processes. PSRs represent state as a set of predictions of future observable events. Because PSRs are defined entirely in terms of observable data, statistically consistent estimates of PSR parameters can be learned effi…
Reservoir computers and RNNs fall short of optimal prediction for stochastic PDFA.
problem Predicting stochastic processes generated by probabilistic deterministic finite-state automata.
method Generalized linear models, Reservoir computers, and Long Short-Term Memory (LSTM) RNNs were tested.
result Each method can fall short of maximal predictive accuracy by up to 50% after training.
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.
PASS model predicts disease progression with both accuracy and interpretability.
problem Balancing accurate disease prediction with clinically interpretable models.
method Phased LSTM units with attention mechanism for non-stationary state dynamics.
result PASS model achieves superior predictive accuracy and interpretable representations.
Study learns state representations from observations for control, proving guarantees.
problem Learning state representations from high-dimensional observations for control.
method Cost-driven approach, learning latent state model to predict costs.
result Proves finite-sample guarantees for near-optimal state representation and controller.
DSSM separates domain-invariant dynamics from domain-specifics in sequential data.
problem Learning cross-domain sequence representations from diverse data domains.
method Introduce disentangled state space models (DSSM) using unsupervised VAE-based training.
result Improves knowledge transfer and robust prediction across domains.
DeepMCP improves CTR prediction by learning better feature representations.
problem Data sparsity in CTR prediction models.
method DeepMCP models user-ad, ad-ad, and feature-CTR relationships through three subnets.
result DeepMCP outperforms state-of-the-art models in CTR prediction.
Enhanced geographical features improve predictive models for colorectal cancer survival curves.
problem Predicting colorectal cancer survival curves in Iowa.
method Used neural networks to explore feature representations, comparing ABC performance.
result Spectral analysis-based representations improve predictive performance by approximately 40%.
EMI uses predictive signals to guide exploration in sparse reward settings.
problem Challenges of reinforcement learning with sparse reward signals.
method Constructs embedding representations of states and actions for forward prediction in the representation space.
result Competitive results on challenging tasks with continuous control and discrete actions.
Study cost-driven state representation learning for control from partial observations.
problem Learning state representation for control from partial and high-dimensional observations.
method Cost-driven state representation learning via predicting cumulative costs.
result Established finite-sample guarantees for near-optimal representation and controller.
Enhances RL agents with predictive internal representations.
problem Improving model-free reinforcement learning agents.
method Introduces Deep InfoMax (DIM) objective to train predictive internal representations.
result Successfully learned predictive representations in synthetic settings.
RNF learns distinct representations for Bayesian filtering steps, improving time series prediction accuracy and uncertainty.
problem Improving time series prediction accuracy and uncertainty using distinct representations for Bayesian filtering steps.
method Introduces Recurrent Neural Filter (RNF) architecture that learns distinct representations for each Bayesian filtering step.
result RNF improves accuracy of one-step-ahead forecasts and provides realistic uncertainty estimates.
The paper proposes an algorithm to learn causal state representations for partially observable environments.
problem Learning task-agnostic state abstractions in partially observable environments.
method The approach involves learning approximate causal state representations from RNNs trained to predict observations given the history.
result The learned state representations are useful for efficient policy learning in reinforcement learning problems with rich observation spaces.
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.
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.
This work learns latent representations to speed up exploration in complex environments.
problem Challenging exploration in high-dimensional state and action spaces with sparse rewards.
method Representation learning using prior experience to learn effective latent representations.
result Learned latent representations reduce the dimensionality of the search space for effective exploration.
Unsupervised neural models predict brain activity better than supervised methods.
problem Understanding how the brain represents visual information without direct supervision.
method Built upon PredNet, used RSA to compare PredNet representations to fMRI and MEG data.
result Unsupervised models trained to predict video frames outperform supervised image classification models in predicting brain activity.
A new method for estimating uncertainty in deep neural networks.
problem Challenges in uncertainty estimation in deep neural networks, especially with increased complexity.
method Decompose tasks into representation learning and state space model for uncertainty estimation.
result The proposed method can estimate predictive distributions on top of existing neural networks.
Persona2vec learns multiple node roles in graphs.
problem Graphs often have nodes with multiple overlapping roles.
method Persona2vec learns multiple node representations based on structural contexts.
result Persona2vec outperforms state-of-the-art models in link prediction.
Self-supervised reward prediction improves RL in sparse reward settings.
problem Data efficiency and sparse reward signals in reinforcement learning.
method Learning a state representation for reward prediction and using it to shape rewards.
result Self-supervised reward prediction enhances RL algorithms in single-goal environments.
A new method for learning controlled dynamical systems efficiently and avoiding local minima.
problem Learning controlled dynamical systems with efficient and robust methods.
method Predictive State Representation with Random Fourier Features (RFFPSR) combining moment-matching, kernel embedding, and local optimization.
result The method avoids local minima and efficiently models controlled dynamical systems.
The paper explores how disentangled representations can improve fairness in prediction tasks.
problem Improving fairness in prediction tasks using disentangled representations.
method Investigates different notions of disentanglement and analyzes representations of state-of-the-art models.
result Disentanglement scores are correlated with increased fairness in prediction tasks.
Method infers causal structure from system behaviors using RKHS and kernel ε-machines.
problem Discovering causal structure in systems with varying external and measurement noise.
method Combines causal states and RKHS for efficient representation and inference of causal structure.
result Robustly estimates causal structure in high-dimensional data with varying noise.
Estimates prediction uncertainty in neural networks using density estimation in representation space.
problem Incorrect predictions with high confidence from models trained on limited data.
method Estimates training data density in representation space and uses it to predict model uncertainty.
result Detects out-of-distribution data without prior exposure, improving model reliability.
Improves PSR learning by refining spectral initialization with PSIM-style updates.
problem Inference performance of PSRs is poor despite good theoretical guarantees.
method Combines spectral algorithms for PSRs with PSIM-style updates for inference-based loss optimization.
result Inference Gradients outperforms PSRs and PSIMs on real and synthetic data.
New DTI model using self-attention molecule representation outperforms state-of-the-art.
problem Predicting drug-target interactions to reduce costs and improve personalized medicine.
method Proposes a new molecule representation using self-attention and a new DTI model.
result Our DTI model outperforms state-of-the-art by up to 4.9% points in precision-recall.
Manifold Mixup improves neural network robustness by interpolating hidden states.
problem Neural networks' incorrect predictions on slightly different test examples.
method Manifold Mixup, a regularizer that encourages less confident predictions on interpolations of hidden representations.
result Neural networks trained with Manifold Mixup learn smoother decision boundaries and fewer directions of variance.
A new trajectory representation method for AI problems.
problem Trajectory prediction and optimization in AI problems.
method Sub-goal trees, recursively partitioning trajectories into sub-segments.
result Sub-goal trees predict trajectories faster and more accurately.
Neural Physicist learns physical dynamics from images.
problem Learning meaningful physical state representations and accurate state transitions from image sequences.
method Neural Physicist uses VAE for state extraction, NP for parameters, and SSM for dynamics.
result Achieves long-term predictions and identifies system degrees of freedom.
Proposes using oracle feedback to predict and prevent errors in RL agents.
problem Mismatches between simulated and real-world environments lead to errors in reinforcement learning.
method Formalizes the problem as a noisy supervised learning task, combines techniques for label aggregation, calibration, and supervised learning.
result Achieves higher predictive performance than baseline methods and can prevent errors by selectively querying an oracle.
End-to-end learnable network for safer self-driving with interpretable intermediate representations.
problem Safe motion planning for self-driving vehicles.
method Differentiable semantic occupancy representation for cost calculation in motion planning.
result Significantly outperforms state-of-the-art planners in imitating human behaviors and producing safer trajectories.
Neural networks improve clinical note representations for predictive modeling.
problem Challenges in using clinical notes for machine learning due to high dimensionality, sparsity, and scarcity of labeled data.
method Used neural networks and transfer learning to learn representations of clinical notes.
result Neural network representations significantly outperformed baseline representations in predictive models.
DySAT learns dynamic graph node representations capturing structural and temporal patterns.
problem Learning latent representations of nodes in dynamic graphs.
method Dynamic Self-Attention Network (DySAT) that combines self-attention layers for structural and temporal dimensions.
result DySAT outperforms state-of-the-art baselines in link prediction on dynamic graphs.
A novel geometric algebra-based KG embedding framework improves link prediction.
problem KG embedding to model entities and relations in a low-dimensional space.
method Utilizes multivector representations and geometric product in geometric algebra.
result Outperforms state-of-the-art models in link prediction experiments.
DeepMDP simplifies complex observations into continuous latent states.
problem Learning from high-dimensional observations in reinforcement learning.
method Trains a DeepMDP model that predicts rewards and next latent states.
result Optimization of DeepMDP objectives ensures quality of latent space and environment model.
A new method predicts dynamical systems better by using two different latent spaces.
problem Predicting the future of dynamical systems with optimal accuracy.
method Uses two different latent mappings for present and future states.
result Optimal 2-mapping method significantly outperforms single latent representation methods.
InfoGraph learns graph-level representations via mutual information maximization.
problem Learning graph-level representations for unsupervised and semi-supervised scenarios.
method Maximizes mutual information between graph-level representation and substructure representations.
result InfoGraph outperforms state-of-the-art methods on graph classification and molecular property prediction.
New insights into state representations in RL help design better learning rules.
problem Lack of automatic feature learning in RL for large or continuous state spaces.
method Bootstrapping methods and theoretical analysis of temporal difference learning.
result State representations differ from other auxiliary-task-based approaches.
This work proposes multiple node representations for graphs, improving link prediction and community analysis.
problem Can nodes be best described by a single vector representation?
method A principled decomposition of the ego-network to learn multiple node representations.
result Improved link prediction accuracy by up to 90% and effective community analysis.
Method learns behavioral states from wearable sensor data.
problem Understanding behavioral patterns from sensor data.
method Non-parametric Bayesian approach to model sensor data.
result Learned behavioral states cluster participants into meaningful groups and predict psychological states.
KINet learns object interactions without supervision for robotic pushing.
problem Lack of supervised data for object-centric forward prediction.
method End-to-end unsupervised framework using keypoint representation and contrastive estimation.
result Automatically generalizes to unseen scenarios and accurately predicts future states.
EvoNet predicts events in time-series data by evolving state graphs.
problem Predicting events in time-series data with interpretable patterns.
method Evolutionary State Graph (ESG) and EvoNet model.
result EvoNet outperforms baselines and provides insights into event predictions.
Multimodal deep learning improves toxicity prediction accuracy.
problem Improving prediction accuracy of chemical compound toxicity.
method Combining multiple neural network types and data representations.
result Significantly better accuracy on a toxicity benchmark.
Improved AutoDML estimator for causal inference using outcome-adapted shared covariate representation.
problem Efficiency in estimating treatment or policy effects in causal inference.
method Outcome-adapted AutoDML estimator that uses a shared covariate representation that is predictive of the outcome but not the Riesz representer.
result Outcome-adapted AutoDML estimator is asymptotically more efficient than baseline AutoDML.
PSRNNs combine RNN and PSR insights for system filtering and prediction.
problem Modeling dynamical systems efficiently and accurately.
method Combines insights from RNNs and PSRs using bilinear transfer functions and tensor decomposition.
result PSRNNs outperform other models in filtering and prediction tasks across multiple datasets.