Multi-expert L2D underfits more severely, requiring new methods.
problem Underfitting in multi-expert L2D settings.
method PiCCE (Pick the Confident and Correct Expert), a surrogate-based method.
result PiCCE effectively reduces multi-expert L2D to a single-expert-like problem, resolving underfitting.
Reinforcement learning mimics expert behavior.
problem Learning from expert demonstrations in reinforcement learning.
method Reduction to reinforcement learning with a stationary reward.
result Expert reward can be recovered and imitation learning is bounded.
Hybrid RL learns from expert state sequences without full action data.
problem Learning from expert state sequences without full action data.
method Tensor-based model to infer unobserved actions; hybrid RL objective.
result Hybrid RL outperforms pure RL and tensor-based action inference.
Expert augmentation improves hybrid model generalization.
problem Limited generalization of hybrid models outside training distribution.
method Introducing expert augmentation to improve hybrid model performance.
result Expert augmentation improves generalization of hybrid models.
ExpCLR uses expert features to improve time-series representation learning.
problem Current representation learning approaches fail to ensure useful properties for time-series data.
method ExpCLR employs expert features to replace data transformations in contrastive learning, ensuring two useful properties for time-series representations.
result ExpCLR outperforms state-of-the-art methods on three real-world time-series datasets.
A model learns new tasks sequentially, using gating autoencoders to select relevant experts.
problem Learning new tasks without storing previous data, ensuring scalability and relevance.
method Sequential learning with gating autoencoders to select relevant experts for new tasks.
result The method selects the most relevant expert for a new task, improving efficiency and scalability.
Adaptive Bayesian learning aggregates experts to improve performance.
problem Bayesian online learning's performance depends on inferential choices.
method Treat Bayesian update rules as experts and aggregate them based on sequential predictive losses.
result The aggregate competes with the best expert in hindsight at a low aggregation cost.
Improved time series forecasting with expert loss integration.
problem Enhancing time series forecasting accuracy and efficiency.
method Adaptive Mixture-of-Experts framework with expert-specific loss integration and online learning.
result Significantly improved forecasting accuracy and computational efficiency.
A new framework estimates expert policy support to create a reward function for imitation learning.
problem Imitation learning from expert trajectories without reinforcement signals.
method Estimating the support of the expert policy to compute a fixed reward function.
result Comparable or better performance than state-of-the-art methods on discrete and continuous domains.
Budgeted deferral framework reduces expert query costs in machine learning.
problem High cost of querying all experts in deferral algorithms.
method Introduces budgeted deferral framework for selective expert querying.
result Empirical results show substantial reduction in training costs without sacrificing accuracy.
Online L2D algorithm for multiclass classification with varying experts.
problem Handling streaming data, changing expert availability, and shifting expert distribution.
method First online L2D algorithm with O ( ( n + n e ) T 2 / 3 ) O((n+n_e)T^{2/3}) O (( n + n e ) T 2/3 ) and O ( ( n + n e ) T ) O((n+n_e)\sqrt{T}) O (( n + n e ) T ) regret guarantees. result Effective extension of standard L2D to settings with varying expert availability and reliability.
HS-MoE selects sparse experts using adaptive priors and data-adaptive gating.
problem Sparse expert selection in mixture-of-experts architectures.
method Combines horseshoe prior with input-dependent gating for data-adaptive sparsity.
result Data-adaptive sparsity in expert usage.
Dynamic partition models learn compact binary representations from data.
problem Learning accurate distributed representations of high-dimensional data.
method The approach involves partitioning variables into expert supports, dynamically adapting partitions based on active experts, and using a smoothed version of the model with separate mixtures for each data dimension.
result Accurate reconstructions of high-dimensional data points achieved with a dozen experts.
Paper proposes consistent estimators for learning to defer decisions to experts.
problem Learning algorithms often ignore expert decision-making in practical scenarios.
method Reduction to cost sensitive learning, novel surrogate loss for consistent estimation.
result Effective approach demonstrated on various tasks, showing consistency.
YRC-Bench benchmarks AI agents learning to collaborate with experts.
problem Learning to recognize when to consult an expert in novel situations.
method Validation strategy and proposer-validator decomposition.
result Insights for future AI safety research.
DMLreg uses expert knowledge to improve model performance in high-dimensional settings.
problem Improving model performance in high-dimensional prediction problems.
method Learning a Mahalanobis distance metric from expert comparisons and integrating it into a regularized linear model.
result DMLreg leads to improvements in model performance when expert knowledge is relevant.
Unified framework for deferring queries to top-k experts, improving accuracy-cost trade-offs.
problem Limitation of existing L2D frameworks to single-expert deferral.
method Top- k k k Learning-to-Defer framework, including adaptive Top- k ( x ) k(x) k ( x ) variant. result Superior accuracy-cost trade-offs with multi-expert deferral.
Improves accuracy and fairness in prediction systems with multiple domain experts.
problem Designing unbiased and accurate deferral systems with multiple experts.
method Proposes a framework for learning a classifier and deferral system that chooses to defer to multiple human experts.
result Significantly improves accuracy and fairness of final predictions compared to baselines.
New model learns better policies from expert demonstrations with higher efficiency.
problem Learning accurate policies from expert demonstrations with high efficiency.
method Generative adversarial imitation learning (GAIL) model that learns f f f -divergence automatically. result Learns better policies with higher data efficiency in physics-based control tasks.
This paper pretrains actor-critic RL algorithms using expert demonstrations.
problem Stability and efficiency of pretraining methods for actor-critic RL algorithms.
method Employ expert demonstrations in actor-critic reinforcement learning framework, ensuring non-global optimal demonstrations are used.
result Our method outperforms RL algorithms without pretraining and is more simulation efficient.
Framework allows systems to defer difficult decisions to unknown experts.
problem Ensuring safety and robustness in autonomous systems with unknown human experts.
method Meta-learning approach that adapts to new experts at test-time using optimization and model-based attention mechanisms.
result Framework can quickly adapt deferral policies to new experts without re-training.
A new method for learning to defer decisions with expert advice improves over standard methods.
problem Learning to defer decisions with expert advice in systems where expert information can be modified after selection.
method An augmented surrogate that operates on the composite expert-advice action space, providing consistency guarantees and excess-risk bounds.
result The method improves over standard Learning-to-Defer and adapts its advice acquisition behavior to the cost regime.
A new method for multi-expert learning-to-defer avoids optimization issues.
problem Optimization issues in multi-expert learning-to-defer systems.
method A decoupled surrogate with a softmax classifier head and independent sigmoid heads per expert.
result First multi-expert L2D guarantee with a constant not growing with the expert pool.
New algorithm reduces expert prediction regret for two experts.
problem Efficient prediction with two experts under fixed time constraints.
method Optimal algorithm based on stochastic calculus techniques.
result Achieves optimal regret of sqrt(T/2π) + O(1) with O(1) per-turn processing time.
SLEEPER combines deep learning with expert rules for accurate sleep staging.
problem Manual sleep staging is tedious and requires expert time.
method SLEEPER uses convolutional neural networks and expert rules to generate interpretable models.
result SLEEPER achieves comparable accuracy to human experts and deep neural networks.
Proposes a mixture of expert architecture for polyhedral classifiers.
problem Learning polyhedral classifiers with high accuracy.
method Uses an expectation maximization algorithm to learn parameters.
result Generalization bounds are derived and the method performs comparably to state-of-the-art approaches.
Paper proposes a method to learn and exceed expert demonstrations in unknown reward environments.
problem Learning to outperform expert demonstrations in unknown reward environments.
method A novel concurrent reward and action policy learning approach with a stereo utility definition.
result The proposed method can outperform expert demonstrations in various environments.
A new algorithm trains experts to safely guide agents in partially observed environments.
problem Existing imitation learning methods for POMDPs can lead to sub-optimal or unsafe policies.
method Derive an objective to encourage the expert to maximize the agent's reward, then use it to train both expert and agent.
result The algorithm produces an expert policy that the agent can safely imitate, outperforming fixed expert policies.
Aggregator learns optimal forecast aggregation in partial evidence environments.
problem Forecast aggregation in repeated binary event settings with expert forecasts.
method Bayesian experts with partial evidence, aggregator learns optimal aggregation.
result Optimal aggregation can be learned in polynomial time for a wide range of partial evidence environments.
This work improves transferability of rewards inferred from expert demonstrations.
problem Transferability of rewards inferred from expert demonstrations under limited access to the expert's policy.
method Proposed principal angles as a measure of similarity and dissimilarity between transition laws. Established sufficient conditions for transferability under limited access.
result Two key results on sufficient conditions for transferability to any and local changes in transition laws.
L2D-CD learns to defer expert recommendations in causal discovery.
problem Combining expert knowledge with data-driven results in causal discovery when expert recommendations may contradict data.
method Adapting learning-to-defer algorithms for pairwise causal discovery, L2D-CD learns a deferral function to select between expert recommendations and data-driven methods.
result L2D-CD outperforms both causal discovery methods and the expert used in isolation, identifying domains where the expert's performance is strong or weak.
EAML combines human and machine learning for better predictions.
problem Limited data and trust in machine learning models.
method Automated method guiding expert knowledge integration into machine models.
result Improved predictions and generalizability with less data.
A mixture of experts model predicts brain activation from word stimuli.
problem Classical encoding models ignore connections among brain regions.
method Mixture of experts capturing ROI-specific brain activity patterns.
result Model predicts entire brain activation with high spatial accuracy.
Meta-algorithm optimizes nonstochastic bandits with infinitely many experts.
problem Maximizing reward by choosing actions sequentially from a set of experts.
method Proposed a variant of Exp4.P for infinitely many experts and a meta-algorithm.
result Proved high-probability upper bound of i l d e O ( i ∗ K + K T ) ilde{\mathcal{O}} \big( i^*K + \sqrt{KT} \big) i l d e O ( i ∗ K + K T ) on regret. Study shows optimal online learning with almost perfect expert advice.
problem Multiclass online learning with advice from up to b experts, focusing on the case where b is small.
method Analyzed the regime where the best expert makes at most b mistakes, showing the optimal forecaster's mistake bound.
result The optimal forecaster's expected number of mistakes is at most log_4(n) + o(log_4(n)) when b = o(log_4(n)).
This paper explores how MoE layers improve deep learning performance.
problem Understanding the Mixture-of-Experts (MoE) layer in deep learning.
method Formal study of MoE layer's effectiveness and mechanism.
result MoE layer improves performance by leveraging cluster structure and non-linearity.
Bayesian models combine experts with a flexible gating mechanism for complex data.
problem Theoretical properties of Bayesian mixture-of-experts models with softmax gating remain unexplored.
method Investigated asymptotic behavior of posterior distribution for density estimation, parameter estimation, and model selection.
result Established posterior contraction rates for density estimation and parameter estimation, providing insights for practical model design.
New method combines adaptive learning rate and flexible prediction for online loss aggregation.
problem Online aggregation of unbounded losses using shifting experts.
method Adapted AdaHedge algorithm with Fixed Share meta-algorithm for signed unbounded losses.
result Improved shifting regret and validity of regret bounds in adversarial setting.
This paper tackles deferral learning with multiple experts, providing strong theoretical guarantees.
problem Optimizing input assignment to experts balancing accuracy and computational cost.
method Introducing new surrogate loss functions and efficient algorithms with strong theoretical learning guarantees.
result Realizable H H H -consistency, H H H -consistency bounds, and Bayes-consistency for deferral learning. DriftMoE adapts to concept drifts in data streams efficiently.
problem Adapting to non-stationary data streams with concept drifts.
method Online Mixture-of-Experts (MoE) architecture with co-training framework.
result DriftMoE achieves competitive results with state-of-the-art stream learning adaptive ensembles.
Framework uses expert intervention to solve long-horizon reinforcement learning tasks.
problem Long horizon robot learning tasks with sparse rewards.
method Option templates and expert intervention to enable high-level task understanding.
result Framework outperforms state-of-the-art approaches by two orders of magnitude.
Optimized deferral improves accuracy in imbalanced settings.
problem Imbalance in expert predictions leads to suboptimal performance in two-stage learning to defer.
method Developed novel cost-sensitive learning algorithms and margin-based loss functions tailored for expert imbalance.
result MILD algorithm shows clear improvements over baselines in image classification and LLM routing tasks.
New method learns internal model from expert demonstrations to estimate rewards.
problem Lack of complete or good action information from expert demonstrations.
method Agent learns internal model of observations from expert state trajectories to estimate rewards.
result Successfully trains good policies from expert game-play videos.
PWIL learns agent behavior from expert using Wasserstein distance.
problem Learning agent behavior from expert demonstrations efficiently.
method Primal Wasserstein Imitation Learning (PWIL) with offline reward function.
result PWIL achieves expert behavior on MuJoCo tasks with minimal interactions.
New framework uses tempered optimism to handle imperfect experts in online learning.
problem Challenges of implicit optimism in practical online learning environments.
method Introduces tempered optimism as a framework for online non-convex learning, modifies existing algorithms.
result Demonstrates tempered optimism as a fruitful paradigm for online non-convex learning.
A new method for ILO with transition model disparity using an intermediary policy.
problem Learning tasks from expert observations with different transition dynamics.
method Training an intermediary policy to match the state transitions of the expert dataset.
result Our method outperforms existing ILO approaches with transition model mismatch.
Paper analyzes approximation and learning of MoEs with (P)ReLU activation.
problem Scaling up deep learning models with MoEs and (P)ReLU activation.
method Approximation and learning-theoretic analysis of MoMLPs with (P)ReLU.
result MoMLPs can uniformly approximate Lipschitz functions with ε \varepsilon ε accuracy using O ( ε − 1 ) \mathcal{O}(\varepsilon^{-1}) O ( ε − 1 ) parameters. Combines neural networks and expert rules for concept-based learning.
problem Extending concept-based learning with machine learning models.
method Form constraints for joint probability distribution and represent feasible set as a convex polytope.
result Neural networks can be trained to satisfy expert rules without violating them.