Library · Machine learning & AI
What Reinforcement Learning got wrong, from 81 theses
Summary written by a model from the records below: Across these studies, reinforcement learning approaches frequently underperformed classical control baselines, suffered from reward hacking, or failed to transfer safely to real-world environments. Researchers repeatedly encountered challenges with training instability, slow convergence, and violations of critical physical constraints during policy deployment. These records come from graduate theses, 2021 to 2026, each labeled with its degree where the repository states it, and each links to its thesis. They were extracted by language models reading the full text, so treat each as a lead to read, not a verdict.
Plus 10 problems the authors left open, listed at the end.
Failure modes, by theses reporting them
Select a mode to jump to its records.
Underperforming classical baselines heuristics and domain-specific controllers
Reinforcement learning policies regularly achieved lower task performance, higher operating costs, or higher energy consumption than proportional-integral-derivative controllers, static portfolios, and greedy heuristics. In several benchmarks, standard algorithms also failed to surpass simple non-learning baselines or alternative supervised formulations.
Tried and failed
reinforcement learning for dynamic resource allocation applied to closed-loop perception control. Outcome: worse than baseline. Reason: RL controller achieved lower detection recall than a simpler model-based PID baseline
Closed Loop Perception for Resource Efficient Autonomous Systems · Georgia Tech · PhD
Tried and failed
history-conditioned adaptive reinforcement learning policies applied to zero-shot sim-to-real robotic manipulation. Outcome: worse than baseline. Reason: adaptive policies conditioned on history transferred worse to the real world than simpler reactive policies
Exploring sim-to-real transfer for learning-based robot manipulation · Imperial · Doctorate
Tried and failed
offline reinforcement learning on pretrained multimodal models applied to embodied robotic policy training. Outcome: worse than baseline. Reason: None
From Multimodal LLMs to Generalist Embodied Agents · Georgia Tech · PhD
Lost to a baseline
Recommender agent trained by deep reinforcement learning demonstrated lower human performance improvement compared to baseline agents exhibiting human-like behavior.
Reciprocal human-machine learning in manufacturing · TU Wien · PhD
Lost to a baseline
Default DQN underperformed A2C and PPO in mean cumulative training reward on the simulated environment.
Reinforcement Learning assisted Adaptive difficulty of Proof of Work (PoW) in Blockchain-enabled Federated Learning · Virginia Tech · Master's
Lost to a baseline
Vanilla PPO without intrinsic rewards failed to achieve higher than 10% success rate on sparse-reward manipulation tasks
Learning to Make Decisions in Robotic Manipulation · MIT · PhD
Tried and failed
model-free reinforcement learning for continuous process control applied to wastewater treatment plant control. Outcome: worse than baseline. Reason: Failed to consistently maintain regulatory compliance across dynamic process operational scenarios.
Reinforcement learning applied to water resource recovery facilities for process optimization and control consolidation · Iowa State · PhD
Tried and failed
reinforcement learning and iterative learning control applied to real-time cellular adaptive control. Outcome: worse than baseline. Reason: outperformed by Bayesian filtering methods for online parameter adaptation
Exploring solutions to the sensing and actuation limitations within cybergenetics · Imperial · PhD
Tried and failed
regression-based Q-function training applied to constrained reinforcement learning. Outcome: worse than baseline. Reason: yielded suboptimal performance compared to distributional RL classification approaches
Risk-Aware Reinforcement Learning with Safety Constraints · MIT · PhD
Tried and failed
tree-structured reinforcement learning applied to branching heuristics in exact symbolic computation. Outcome: worse than baseline. Reason: Failed to outperform standard hand-crafted variable selection heuristics
Reinforcement Learning in Buchberger's Algorithm · Cornell · PhD
Tried and failed
on-policy reinforcement learning with clipped surrogate objective applied to continuous control benchmark tasks. Outcome: worse than baseline. Reason: failed to learn effectively and showed poor performance compared to off-policy reinforcement learning
Robustness of Reinforcement Learning Systems in Real-World Environments · MIT · PhD
Lost to a baseline
Learned RL policy for sepsis management was inferior in true expected reward to the behavior policy despite outperforming it on Weighted Importance Sampling (WIS) and Model-Based off-policy evaluation.
Towards Rigorously Tested & Reliable Machine Learning for Health · MIT · PhD
Lost to a baseline
Under NEWS2 reward, the naive weighted random baseline beat all offline RL and SL policies on WIS (-3.78) and WISt (-3.78) on the full sepsis test set.
Dynamic treatment regime for electronic health record · Oxford · PhD
Lost to a baseline
R-learning models (RL1/RL2) that maximize average reward rate were outperformed in WAIC and protected exceedance probability by exponential temporal discounting Q-learning (QL2).
Adaptive time-dependent decisions: behavior, cognition, and neural mechanisms · Boston University · PhD
Lost to a baseline
Uniform Sampling achieved 95.8% on Xstest Should Respond compared to 91.6% by 1K 5-Rules RA on HH-RLHF reward model training
Problems in High-Dimensional Estimation and Large Language Models · Harvard · PhD
Tried and failed
high automation of routines via reinforcement learning applied to decision making in dynamic environments. Outcome: worse than baseline. Reason: lock-in to suboptimal paths reduces exploratory learning and slows adaptation to environment changes
Tried and failed
Adaptive reinforcement learning with dynamic policy shaping applied to interactive game policy learning. Outcome: worse than baseline. Reason: uncompensated initial estimation errors degraded performance under negative feedback
Learning to Teach: Models for Semantic and Adaptive Personalization of AI Tutors · MIT · PhD
Tried and failed
value-aware model learning without target value updates applied to continuous control model-based reinforcement learning. Outcome: worse than baseline. Reason: optimizing against stale value function estimates degrades dynamic model accuracy below standard maximum likelihood estimation
Leveraging Value-awareness for Online and Offline Model-based Reinforcement Learning · Georgia Tech · PhD
Lost to a baseline
Proximal Policy Optimization (PPO) reinforcement learning performed less stably at later iterations than the simpler greedy search heuristic.
Optimizing Brain Stimulation for Parkinson's Disease, Memory Enhancement, and Optogenetic Control · Georgia Tech · PhD
Lost to a baseline
Model-free PPO was beaten by the static Markowitz portfolio across substantial portions of the cumulative reward empirical CDF under alpha decay dynamics.
Reinforcement learning for sequential decision-making: a data driven approach for finance · Scuola Normale Superiore · PhD
Considered and rejected
Considered and rejected: Rejected policy gradient / reinforcement learning (REINFORCE) for policy optimization due to higher complexity, training instability, and inferior performance compared to Gumbel-Softmax.
Efficient deep learning for image and video understanding · Boston University · PhD
Tried and failed
policy gradient reinforcement learning with reranking reward applied to dialogue response generation. Outcome: worse than baseline. Reason: optimizing reranking rewards via policy gradient produced shorter and less informative outputs than test-time reranking
Specific response generation and elaborative discourse structure · UT Austin · PhD
Lost to a baseline
When temporal logic advice was completely perfect, reward shaping was outperformed in initial learning speed and reward by hard action shielding.
Towards long-term deployment of general-purpose service robots · UT Austin · PhD
Lost to a baseline
DDPG + CAPS achieved lower reward (181.98 ± 87.18) on LunarLanderContinuous-v2 than unregularized DDPG (217.04 ± 51.61)
Conditioning behavior styles of Reinforcement Learning policies · Boston University · PhD
Tried and failed
reinforcement learning for discrete sequential optimization applied to power grid unit commitment. Outcome: worse than baseline. Reason: RL operating costs were 2x to 13x higher with severe constraint violation and load shedding
Secure and cost-effective operation of low carbon power systems under multiple uncertainties · Imperial · PhD
Tried and failed
model-free reinforcement learning control applied to commercial refrigeration energy management. Outcome: worse than baseline. Reason: agents struggled across variable operating horizons and dynamic price profiles compared to simple PI control
Design and deployment of data-driven control retrofits for energy systems in commercial buildings · Imperial · PhD
Tried and failed
increasing discount factor and prediction horizon applied to reinforcement learning with model predictive control. Outcome: worse than baseline. Reason: longer horizons amplified the effect of high system stochasticity, degrading controller performance
Human-in-the-loop energy-efficient building HVAC control · Georgia Tech · PhD
Tried and failed
discrete value-based reinforcement learning applied to process aeration control. Outcome: worse than baseline. Reason: agents resulted in significantly higher energy consumption than standard manual control
Reinforcement learning applied to water resource recovery facilities for process optimization and control consolidation · Iowa State · PhD
Lost to a baseline
Hand-tuned MLP continuous control policies achieved >3000 reward on Centipede-20, beating unfrozen GNN NerveNet which was capped near 1600 reward.
A minimalist approach to deep multi-task learning · Oxford · PhD
Reward specification defects reward hacking and misaligned objectives
Policies optimized proxy rewards by exploiting format compliance or non-target subtask rewards without actually solving the underlying tasks. Unregularized, sparse, or misspecified reward formulations also produced repetitive degenerative language outputs, endless hovering behavior, or training collapse.
Tried and failed
potential-based reward shaping in reinforcement learning applied to robot manipulation and locomotion benchmarks. Outcome: worse than baseline. Reason: failed to outperform heuristic-only policies under finite data regimes
Generative Discovery via Reinforcement Learning · MIT · PhD
Tried and failed
using temporal logic robustness as reinforcement learning reward applied to robot control with temporal logic specifications. Outcome: worse than baseline. Reason: robustness scores alone provide poor reward landscape and sparse feedback for reinforcement learning
Right on Time: Structured Robot Learning of Planning and Control Strategies for Signal Temporal Logic · MIT · PhD
Considered and rejected
Considered and rejected: Rejected pure Reinforcement Learning for multi-turn code generation, choosing imitation learning via one-step recoverable MDP reduction to avoid sparse reward exploration
Reasoning in the Wild · Cornell · PhD
Considered and rejected
Considered and rejected: Rejected using standard TD error or reward shaping (RS) directly for trajectory reward modification in CLTV, as empirical ablation showed the proposed TS formulation achieved superior transfer scores.
Learning in Non-Stationary Environments · Leibniz Universität Hannover · PhD
Tried and failed
reinforcement learning with custom reward function applied to energy management optimization. Outcome: did not converge. Reason: downward trend in reward function over training time due to poor learning or unsuitable reward design
Probabilistic Approaches to Enhance Safety and Energy Management of Energy Transition Components and Systems · Texas Tech · PhD
Tried and failed
fine-tuning reinforcement learning with simplified reward function applied to underactuated dynamic swing-up control. Reason: switching from hybrid to pure time-based reward provided no performance improvement over pre-trained policy
Optimal control and reinforcement learning in simple physical systems · Harvard · PhD
Tried and failed
reward shaping in reinforcement learning applied to thermal dynamics control. Outcome: did not converge. Reason: configurations either failed to converge or resulted in significantly slower training compared to gradient modification baselines
Fusing Pre-existing Knowledge and Machine Learning for Enhanced Building Thermal Modeling and Control · EPFL · PhD
Tried and failed
pure reinforcement learning and reward matching applied to human pedagogical feedback modeling. Outcome: did not generalise. Reason: monotonic reward gradients failed to capture non-monotonic human evaluative teaching patterns
The cognitive contours of punishment · Harvard · PhD
Tried and failed
reinforcement learning with single task reward classifier applied to language model generation. Reason: optimizing solely for task reward caused repetitive degenerative outputs without fluency regularization
Improving Data Efficiency and Availability for Alignment · Cornell · PhD
Tried and failed
reinforcement learning with domain-specific reward only applied to conditional text generation. Outcome: worse than baseline. Reason: optimizing solely for domain-specific metrics degraded natural language fluency and readability
Effective Modeling in Medical Imaging with Constrained Data · MIT · PhD
Tried and failed
LLM zero-shot reward function code generation applied to dexterous robotic hand manipulation tasks. Reason: Generated reward code failed to execute consistently across multiple random seeds.
Personalized, Safe, and Interactive Robot Programming via Human Demonstrations · Georgia Tech · PhD
Tried and failed
reinforcement learning with misconfigured reward mechanism applied to traffic simulation control. Outcome: no signal. Reason: improper reward function prevented policy quality differentiation, causing persistent random exploration without learning
Bridging the gap: unifying transportation planning and operations through enhanced travel demand modeling · UT Austin · PhD
Tried and failed
adversarial inverse reinforcement learning with positive rewards applied to episodic goal-reaching tasks. Reason: positive rewards near the target incentivized hovering indefinitely rather than terminating the episode
Estimation and control of visitation distributions for reinforcement learning · UT Austin · PhD
Tried and failed
human feedback shaping with low-accuracy error signals applied to reinforcement learning convergence acceleration. Outcome: worse than baseline. Reason: decoding accuracy below threshold injected noise that confused the agent rather than guiding policy optimization
ON THE INTERPLAY BETWEEN BRAIN-COMPUTER INTERFACES AND MACHINE LEARNING ALGORITHMS: A SYSTEMS PERSPECTIVE · Georgia Tech · PhD
Tried and failed
linear reward-inaction reinforcement learning applied to learning probabilistic target distributions. Reason: algorithms exhibit probability-maximizing behavior rather than matching mixed/variable target distributions
Learning Grammar Distributions with Limited Feedback · Penn · PhD
Considered and rejected
Considered and rejected: Rejected omitting reward hacking of -1 for non-target actions during targeted DRL backdoor attacks, as it prevents the agent from learning the backdoor behavior.
Enhancing deep learning security through explainability and robustness · Boston University · PhD
Considered and rejected
Considered and rejected: Rejected manually assigning constant rewards to door-opening behaviors in KeyCorridor because it causes reward hacking (repeatedly opening/closing doors).
Specification-guided imitation learning · Boston University · PhD
Considered and rejected
Considered and rejected: Rejected equal delayed reward assignment across all MDP steps in text classification DRL because it ignores step-wise token certainty.
Understanding and Combating Online Social Deception · Virginia Tech · Doctorate
Considered and rejected
Considered and rejected: Using raw task rewards in multitask NetHack, rejected because frequent Scout rewards caused the agent to ignore Score and Gold tasks
Reading to Learn · University of Washington · PhD
Considered and rejected
Considered and rejected: Rejected labeling dialogue data with learned reward model in favor of +-1 binary labels due to RM noise and inaccuracy
Towards a Theory and Practice of Open-Ended Reasoning with Generative Models · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected using learned reward models instead of value functions for state abstraction; value functions are smoother and capture long-term environmental effects.
Transfer in sequential decision making · Oxford · PhD
Generalization failures sim-to-real transfer gaps and sensitivity to distribution shifts
Policies trained in simulated settings failed to transfer effectively to physical robotic hardware due to unmodeled dynamics and uncaptured human biomechanical variations. Agents also experienced severe degradation when exposed to out-of-distribution demand surges, sensor faults, high-dimensional raw features, or unseen task benchmarks.
Tried and failed
direct sim-to-real transfer of reinforcement learning policies applied to embodied robotic navigation. Outcome: did not generalise. Reason: unmodeled real-world physical dynamics mismatch between the simulation environment and physical hardware
4D audio-visual learning: a visual perspective of sound propagation and production · UT Austin · PhD
Tried and failed
training reward optimization in reinforcement learning applied to real-world robot deployment. Outcome: did not generalise. Reason: Empirical training reward showed negative correlation with physical robot performance.
Sampling-based optimized adaptive discretization and its applications in robotics · Iowa State · Master's
Tried and failed
sim-to-real reinforcement learning without human domain randomization applied to physical human-robot interaction policies. Outcome: did not generalise. Reason: simulated biomechanical models failed to capture the complexity and variability of real human biomechanics
Robotic Caregivers -- Simulation and Capacitive Servoing for Physical Human-Robot Interaction · Georgia Tech · PhD
Tried and failed
regularizing reinforcement learning against a teacher policy applied to robot manipulation skill transfer. Outcome: did not generalise. Reason: biased initialization degraded downstream transfer performance on difficult tasks
Learning Motion Policies for Dexterous Manipulation with Geometric Fabrics · Georgia Tech · PhD
Tried and failed
classical POMDP inverse reinforcement learning algorithms applied to reward learning under misspecified dynamics. Outcome: worse than baseline. Reason: algorithms suffer high estimation error when the forward agent's model dynamics or observation likelihoods are misspecified
Inverse Reinforcement Learning: A Microeconomics-Based Approach · Cornell · PhD
Tried and failed
reinforcement learning process control under sensor failure applied to wastewater treatment process setpoint optimization. Outcome: did not generalise. Reason: agents exposed to simulated sensor faults failed to maintain effluent constraints due to lack of fault robustness
Reinforcement learning applied to water resource recovery facilities for process optimization and control consolidation · Iowa State · PhD
Tried and failed
information-regularized actor-critic reinforcement learning applied to predicting human performance in unchunked states. Outcome: did not generalise. Reason: fitted policy complexity penalties failed to robustly predict empirical performance benefits across task states
Policy compression: Acting with limited cognitive resources · Harvard · PhD
Tried and failed
deep reinforcement learning for continuous setpoint control applied to wastewater treatment process control. Outcome: did not generalise. Reason: agents failed under elevated influent flow stress tests due to clarifier overloading violating effluent limits
Reinforcement learning applied to water resource recovery facilities for process optimization and control consolidation · Iowa State · PhD
Tried and failed
reinforcement learning with strict zero slack time applied to real-time dynamic bipartite matching. Outcome: did not generalise. Reason: marginal performance gains over greedy baseline and poor zero-shot scale transferability
Integrating Econometric Behavioral Models into Transportation Network Optimization · Cornell · PhD
Tried and failed
reinforcement learning on high-dimensional raw feature space applied to portfolio management and stock selection. Outcome: overfit. Reason: extremely high-dimensional state space with raw tabular features caused severe overfitting to training data
Next-Generation Intelligent Portfolio Management · MIT · PhD
Tried and failed
reinforcement learning for compiler pass ordering applied to code optimization sequences. Outcome: did not generalise. Reason: optimization sequences learned from random programs overfitted and degraded performance on unseen benchmarks
Supercharging Programming through Compiler Technology · MIT · PhD
Tried and failed
unregularized reinforcement learning applied to human-agent multi-agent coordination. Outcome: did not generalise. Reason: agents diverged from human play styles, reducing action prediction accuracy and cooperation
Building Strategic AI Agents for Human-centric Multi-agent Systems · MIT · PhD
Tried and failed
pretrained language model next sentence prediction applied to text-based reinforcement learning environments. Outcome: did not generalise. Reason: failed to capture commonsense relationships needed for sequential decision making
Infusing structured knowledge priors in neural models for sample-efficient symbolic reasoning · EPFL · PhD
Tried and failed
longer context windows and return discounting in DT applied to offline reinforcement learning trajectory generation. Reason: did not improve goal completion percentage or cumulative reward performance over baseline
Leveraging offline data for safe and efficient autonomous navigation in multi-agent environments · Iowa State · Master's
Lost to a baseline
In simulation, the MARL policy without domain randomisation achieved higher average reward than the medium domain randomisation policy
Collaborative and safe autonomous driving through multi-agent deep reinforcement learning · Imperial · PhD
Lost to a baseline
Customized vehicle automation using expert-coded rewards achieved 61.67% prediction accuracy on unobserved neutral trips, losing to the non-customized baseline of 65.33%
Toward Trust-calibrated Customized Vehicle Automation · University of Washington · PhD
Considered and rejected
Considered and rejected: Avoided evaluating QA-Feedback ChatGPT generations with trained reward models because T5-trained classifiers fail to generalize out-of-distribution.
Human-Centered Interactive Information Seeking · University of Washington · PhD
Tried and failed
reinforcement learning integrated with integer linear programming applied to dynamic matching under demand surges. Outcome: did not generalise. Reason: performance declined and was worse than baseline during out-of-distribution transfer under extreme demand
Integrating Econometric Behavioral Models into Transportation Network Optimization · Cornell · PhD
Tried and failed
reinforcement learning fine-tuning with transfer learning applied to power distribution network design optimization. Outcome: did not converge. Reason: large shifts in the target performance boundary rendered pre-trained representations insufficient for optimization convergence
Design Optimization of Power Delivery Networks in Packaging · Georgia Tech · PhD
Training instability slow convergence and high sample overhead
Policy training frequently suffered from high variance, value overestimation bias, and optimization collapse caused by noisy approximations or negative entropy coefficients. Algorithms were also set aside because of excessive data requirements, long convergence times exceeding one thousand iterations, and extensive hyperparameter tuning demands.
Considered and rejected
Considered and rejected: Decided against Reinforcement Learning (DQN, PPO, SAC) due to lack of environment feedback, inability to handle expanding action spaces, and excessive data requirements.
Robust machine learning methods for high-dimensional datasets with applications in genomics and finance · Imperial · PhD
Tried and failed
reinforcement learning with negative entropy coefficient applied to reinforcement learning policy optimization. Outcome: unstable. Reason: a negative entropy coefficient severely penalized exploration, leading to training collapse and dropped returns
Computational Simulation and Machine Learning for Quality Improvement in Composites Assembly · Virginia Tech · Doctorate
Tried and failed
on-policy reinforcement learning with stochastic exploration applied to reservoir history matching. Outcome: too slow. Reason: stochastic exploration caused the agent to take suboptimal actions, wasting significant compute before recovery
Machine Learning Based Algorithms for Improving Forecasting in Subsurface Energy Resources · MIT · PhD
Considered and rejected
Considered and rejected: Rejected multi-objective RLHF (Reinforcement Learning from Human Feedback) due to training instability, extensive hyperparameter tuning requirements, and computational inefficiency compared to MODPO.
Essays on Digital Content Strategies: Creation, Diffusion, and Monetization · Harvard · PhD
Considered and rejected
Considered and rejected: Rejected Reinforcement Learning (RL) agents for decision modeling because they incur training overhead and require problem-specific DNN architectures compared to search algorithms.
Network Digital Twins: A Paradigm for Network Management Applications · Carleton University · PhD
Considered and rejected
Considered and rejected: Rejected Policy Gradient reinforcement learning due to slow convergence requiring over 1,000 iterations to reach cumulative reward criteria
Preventing Breaks in Embodiment in Immersive Virtual Reality · EPFL · PhD
Tried and failed
deep Q-learning for interactive control applied to human-robot social interaction. Outcome: did not converge. Reason: insufficient training episodes and unmodeled human response latency causing reward misclassification
Designing Socio-Spatial Interfaces for Embodied and Situated Social Interaction · Cornell · PhD
Tried and failed
Deep Deterministic Policy Gradient reinforcement learning applied to continuous control in power systems. Outcome: unstable. Reason: Q-value overestimation bias leading to degraded policy performance and training instability
Physics Informed Learning-Based Frequency Regulation and Virtual Inertia Control for Renewable Energy Generators in Power Systems · Carleton University · PhD
Considered and rejected
Considered and rejected: Rejected DDPG reinforcement learning policy in favor of PPO due to PPO's superior performance, faster convergence, and fewer tunable hyperparameters.
Topology-aware efficient and transferable model compression using graph representation and reinforcement learning · Iowa State · Master's
Failure to enforce physical safety legal rules and operational constraints
Unconstrained reward optimization caused severe physical capacity, branch flow, and thermal line limit violations in power distribution networks. Practitioners also rejected real-world exploration and unconstrained policies due to the risk of physical hardware damage and failures to obey legal rules.
Lost to a baseline
On the IEEE 14-bus benchmark, reinforcement learning (RL) achieved a lower optimality gap deviation from MIPS (0.6%) than the proposed ICNN (1.53 ± 0.27%), albeit violating feasibility constraints.
Data-driven Methods for Optimal Power Flow in Smart Grids · York University · PhD
Considered and rejected
Considered and rejected: Reinforcement learning in the real-world, rejected due to risk of equipment damage from required negative/unsuccessful exploration trails
Perception Based UAV Path Planning for Fruit Harvesting · Johns Hopkins · Master's
Considered and rejected
Considered and rejected: Rejected purely offline reinforcement learning / regression-based policy evaluation due to confounding and lack of penalization for suboptimal training decisions
Essays on Digital Transformation and Business Decision-Making · University of Washington · PhD
Considered and rejected
Considered and rejected: Full RL-based shared reward for formation control; rejected as its navigation accuracy and formation robustness were insufficient compared to hybrid position-based control.
Cooperative Payload Transportation by UAVs: A Model-Based Deep Reinforcement Learning (MBDRL) Application · Virginia Tech · Master's
Considered and rejected
Considered and rejected: Rejected relying solely on implicit RL or standard PPO reward shaping (distance travelled/collision avoidance) because it failed to enforce legal compliance such as stopping at red lights and yielding to emergency vehicles
Enhancing autonomous vehicle decision-making through scenario-based traffic rule integration · Imperial · Master's
Considered and rejected
Considered and rejected: Standard Reinforcement Learning / MDP methods rejected because they cannot achieve better than O(sqrt(T)) regret and fail to exploit queueing/extreme-point network structure.
Learning-NUM: Utility Maximization in Stochastic Queueing Networks · MIT · PhD
Tried and failed
unconstrained reinforcement learning for constrained control applied to power distribution network optimization. Reason: optimizing reward without constraint enforcement causes severe physical capacity and thermal line limit violations
Multi-agent optimization and learning methods for sustainable, smart and resilient power distribution systems and microgrids · Iowa State · PhD
Left open by the authors
Problems the authors named and did not get to.
Left open
Learn neuro-symbolic low-level robotic skills using reinforcement learning instead of relying on expert demonstrations. Blocker: Lacks specific RL algorithm choices, reward formulations, and environment setups beyond general intent
Neuro-Symbolic Learning for Bilevel Robot Planning · MIT · PhD
Left open
Apply reinforcement learning fine-tuning on pre-trained imitation robot policies to achieve over ninety percent real-world deployment success rates. Blocker: Requires real-world physical robot hardware and physical experimental environments for deployment and evaluation.
Scaling robot learning with heterogeneous data from the real world, simulation, and the web · UT Austin · PhD
Left open
Deploy the language-guided reinforcement learning and adaptation models on physical robots using sim-to-real transfer and safe RL techniques. Blocker: Requires access to physical robotic hardware.
Using natural language to aid task specification in sequential decision making problems · UT Austin · PhD
Left open
Train graph neural networks using reinforcement learning for decentralized robot control instead of oracle imitation learning. Blocker: None
Machine Learning On Large-Scale Graphs · Penn · PhD
Left open
Transition FiLM-Nav to end-to-end reinforcement learning and safe real-world fine-tuning on robotic hardware. Blocker: Requires physical robot hardware and real-world testing environments for safe real-world fine-tuning
From Web to World: Harnessing Foundation Models for Intelligent Robotic Assistants in Real-World Environments · Georgia Tech · PhD
Left open
Deploy and evaluate unsupervised model-based reinforcement learning agents on physical, real-world robotic systems to assess scalability. Blocker: Requires access to physical robotic hardware and real-world testing environments.
LEARNING TO ACT FROM DIVERSE DATA SOURCES VIA WORLD MODELS · Penn · PhD
Left open
Transfer learned reinforcement learning motion planning policies across varying robot kinematic chains and dynamic environments in simulation. Blocker: None
Estudio de la planificación de trayectorias en entornos dinámicos para manipuladores industriales mediante aprendizaje por refuerzo · University of Deusto · PhD
Left open
Deploy reinforcement learning policies safely onto physical collaborative robots in real-world industrial production environments under latency constraints. Blocker: Requires physical collaborative robot hardware and an industrial production testing environment.
Aprendizaje por refuerzo en robótica colaborativa: un estudio sobre la influencia de parámetros humanos en el desarrollo de habilidades robóticas · University of Deusto · PhD
Left open
Train reward models and apply RLHF to align a financial LLM's sentiment outputs for portfolio management. Blocker: None
Next-Generation Intelligent Portfolio Management · MIT · PhD
Left open
Investigate regularization-based methods to incorporate prior financial knowledge into reinforcement learning policy training for portfolio execution. Blocker: No specific regularization formulations, loss terms, or financial prior targets are detailed beyond a broad research direction
Reinforcement learning for sequential decision-making: a data driven approach for finance · Scuola Normale Superiore · PhD
Who has worked on this
Authors of the public theses cited on this page, most-cited first.
- Henry Croll · Iowa State · PhD · 2024
Reinforcement learning applied to water resource recovery facilities for process optimization and control consolidation - Youngseo Kim · Cornell · PhD · 2025
Integrating Econometric Behavioral Models into Transportation Network Optimization - Zijie Zhao · MIT · PhD · 2024
Next-Generation Intelligent Portfolio Management - Alessio BRINI · Scuola Normale Superiore · PhD · 2022
Reinforcement learning for sequential decision-making: a data driven approach for finance - MENGJIE CHENG · Harvard · PhD · 2026
Essays on Digital Content Strategies: Creation, Diffusion, and Monetization - Yue Meng · MIT · PhD · 2026
Right on Time: Structured Robot Learning of Planning and Control Strategies for Signal Temporal Logic - Xiaomin Li · Harvard · PhD · 2026
Problems in High-Dimensional Estimation and Large Language Models - Asier González Santocildes · University of Deusto · PhD · 2026
Aprendizaje por refuerzo en robótica colaborativa: un estudio sobre la influencia de parámetros humanos en el desarrollo de habilidades robóticas - Benjamin Smart · Imperial · PhD · 2025
Exploring solutions to the sensing and actuation limitations within cybergenetics - Ishaan Grover · MIT · PhD · 2025
Learning to Teach: Models for Semantic and Adaptive Personalization of AI Tutors - Alexander Havrilla · Georgia Tech · PhD · 2025
Towards a Theory and Practice of Open-Ended Reasoning with Generative Models - Serena Guo · Cornell · PhD · 2025
Designing Socio-Spatial Interfaces for Embodied and Situated Social Interaction
Checking a claim in this area?
We can run the same search on any method or claim. If nothing turns up, we will say so, and that proves nothing on its own.