Papers where the original authors have engaged with our plain-language explanation.
For every paper on this page, at least one of the original authors has seen our plain-language explanation and engaged with it — either confirming it reads accurately or requesting corrections that we then applied. An endorsement does not mean the authors formally approve every sentence, but it does mean the explanation has passed the eyes of the people who wrote the paper.
949 papers reviewed by authors · 311–320 / 949
This paper establishes a theoretical link between coherent information and neural network loss to develop a transformer-based decoder that accurately predicts quantum error correction thresholds and significantly outperforms traditional minimum weight perfect matching decoders, while also proving the optimality of a novel soft post-selection scheme.
This study presents a comprehensive benchmark of eight machine learning and deep learning models for predicting wound-rotor synchronous motor performance, demonstrating that neural-network-based architectures, particularly the FT-Transformer, significantly outperform tree-based models in accuracy and computational speed while offering a robust multi-seed reproducibility protocol and Pareto analysis of cost-accuracy trade-offs.
By combining new JWST observations with archival data to detect high abundances of refractory SiO in WASP-121 b's atmosphere, this study reveals that the ultra-hot Jupiter's composition resulted from accretion from multiple reservoirs and its current high-obliquity orbit can be explained by post-formation dynamical events.
This paper analyzes three decades of multiwavelength variability in the BL Lac object PKS 0735+178, associated with a high-energy neutrino event, revealing correlated emission with frequency-dependent delays and a ~12-day optical-gamma lag that support a jet propagation model driven by shocks and base energy variations rather than jet precession.
This paper introduces a Bayesian neural network implementation within the aenet-PyTorch framework and systematically compares its uncertainty quantification capabilities against deep ensembles on TiO structures to guide the development of reliable machine learning interatomic potentials.
This paper challenges the assumption that low mutual information between inputs and representations implies geometric compression, revealing instead a complex, nonlinear relationship where generalization acts as a confounding factor rather than a direct consequence.
The paper introduces BabelJudge, an open-source benchmark and reliability audit framework that evaluates LLM-as-a-Judge models across multiple languages and agent trajectories by generating gold-standard labels through controlled perturbations to expose hidden failure modes like position and verbosity biases that raw accuracy metrics fail to capture.
This paper investigates the similarities and differences between LLM-generated and human-written stories by applying narratological definitions to analyze eight intricate dimensions of character portrayal, aiming to determine if AI models produce characters with comparable variety and depth to human authors.
This paper demonstrates that compact, task-adapted small language models (under 3B parameters) can significantly outperform zero-shot frontier LLMs in both general and literary relation extraction when trained on specific domain data, offering a hardware-efficient and private alternative without requiring generative scaling.
This paper demonstrates that externally launched helicon waves can effectively suppress runaway electron growth in the DIII-D tokamak through resonant pitch-angle scattering in both ideal and non-ideal antenna configurations, while highlighting that anomalous resonance enhances runaway populations and significant propagation challenges remain for post-disruption applications.