reinforcement learning

Why Transformers Cannot Invent: The Manifold Confinement Problem and the Thermodynamic Engine of Scientific Discovery

The contemporary artificial intelligence research consensus posits that scaling test-time compute, Monte Carlo tree search, and Reinforcement Learning with Verifiable Rewards (RLVR / GRPO) will inevitably culminate in automated, superhuman scientific discovery. In this paper, we demonstrate the structural impossibility...

Data for "Auditing Single-Agent Reinforcement Learning for EV Charging Assignment: A Protocol-Amended Comparison of Trained, Untrained, and Heuristic Policies"

Version 2 of the data deposit for the manuscript "Auditing Single-Agent Reinforcement Learning for EV Charging Assignment: A Protocol-Amended Comparison of Trained, Untrained, and Heuristic Policies" (Algorithms, MDPI, manuscript ID algorithms-4572374). Version 2 replaces the preliminary data-only content of version 1...

Diagnosing Estimator Homomorphism Compatibility in Reinforcement Learning with Verifiable Rewards

Reinforcement learning with verifiable rewards (RLVR) evaluates decoded outcomes, whereas policy-gradient estimators operate on concrete trajectories and batch-dependent training state. These two levels need not respect the same equivalence relation. We formalize estimator homomorphism compatibility: when several concr...

Spectral SU(3) Gauge Invariant on Four-Dimensional Euclidean Lattices: Wilson Loop Expectation Values, Area Law Confinement, and the Non-Perturbative Mass Gap

Complete lattice computation of Wilson loop expectation values under SU(3) gauge theory on a four-dimensional Euclidean lattice. Establishes color confinement via the area law and extracts the non-perturbative Yang-Mills mass gap of 77.71 MeV from the two-point glueball correlator spectral decomposition. ⚖️ PROPRIETARY...

Nonlinear Inversion of Dissipative Trajectories in Hilbert Spaces via Riccati Differential Matrix Control and Saddle-Node Bifurcation Stabilization

Rigorous formulation and solution of the deterministic nonlinear trajectory inversion problem in four-dimensional Hilbert spaces via the Riccati Differential Equation (DRE) and Continuous Algebraic Riccati Equation (CARE). Derives the optimal LQR feedback law, proves entropy neutralization, and establishes bifurcation...

QTrust-PPO: Confidence-Aware Finite-Shot Policy Optimization for Variational Quantum Reinforcement Learning

Initial public release of QTrust-PPO. This release contains the implementation notebooks, processed statistical results, figures, and table data for the QTrust-PPO experiments. Included components: Primary three-environment benchmark Confidence and rollback ablations Output-noise robustness experiments VQC depth analys...

A Review of Machine Learning-Based Theft Detection for Smart Energy Meters

A literature review of Machine Learning applications in the domain of Electricity Theft Detection (ETD) that is grounded in 71 scholarly studies produced between 2021 and 2026 and selected according to criteria of machine learning application in the process of theft/NTL detection and accompanying empirical validation.

Preference Optimization for Open-Weight Language Models: A Survey and Low-Resource RLHF Case Study

Open-weight language-model post-training combines preference learning, reinforcement learning, verifiable feedback, and increasingly, agentic training and on-policy distillation. Comparing these systems requires separating their optimization algorithms from their feedback sources, sampling procedures, and evaluation ta...

Robotic Garment Folding via VLA Policy: Engineering, Not Pure Math — E8 Intelligence Research

FINDING: The search results are video titles/abstracts, not actual solutions. No mathematical content is extractable. The only concrete item is an arXiv abstract on robotic garment folding (LeHome Challenge 2026), which is engineering, not pure mathematics. MATH: No equations, constants, or ratios are present in the pr...

Towards an Integrated Production Scheduling and Maintenance Planning Using Reinforcement Learning

An improved MADDPG-based framework for bidirectional safety management of construction robot interaction

Purpose Safety risks are particularly pronounced when multiple construction robots collaborate in complex construction environments. To enhance collaborative efficiency while ensuring operational safety, this study develops and evaluates, in a scaled simulation, a bidirectional safety management approach that balances...

A battery life prediction method based on graph neural network assisted by deep reinforcement learning

With the wide utilization of new energy batteries, the accurate prediction of battery life has become an urgent technical issue that needs to be addressed. Studies have developed battery life prediction models based on correlation analysis for feature extraction. This study introduces a DQN-based graph structure genera...