The Archive

Framework using causal reasoning to distinguish genuine individual improvement from strategic gaming in algorithmic recourse systems.

Structured extraction of event type, impact scope, and temporal horizon from financial news using LLaMA-3.1-70B outperforms sentiment-only prediction.

Diagnostic analysis of KV cache precision effects on token prefix reconstruction in Qwen2.5-derived language model inference.

Multi-agent reinforcement learning approach coupling assortment, sourcing, and routing decisions for end-to-end supply chain optimization.

Do you remember Friend? The Friend that launched an AI pendant, spent $1.8 million of its $2.5 million to acquire friend.com, and plastered the NYC subway with ads promoting artificial companionship? Yeah well, if you didn't remember, Friend is back. And now, it's twice the price. Introducing friend. pic.twitter.com/T5vUCj7iqv - friend (@friend) July 30, 2026 This morning, Friend launched a new ad showing two people talking to their Friend pendants about very personal life problems. It didn't say it explicitly, but the ad showcased a new feature the pendant now includes: a speaker. Before, Fr...

Neuro-symbolic framework combining Swin Transformer vision with fuzzy rule-based reasoning for explainable sewer pipe severity classification.

LLMs exhibit salience bias, prioritizing explicit numerical distractors over implicit commonsense knowledge in reasoning tasks.

Spanish domain-adapted mental health screening models with incremental context expansion for early risk detection from social media.

On-board inference and generative data augmentation reduce downlink bottlenecks for aircraft detection from nanosatellites.

EndoCLIP, a vision-language foundation model trained on 280k colonoscopy reports, improves lesion detection and report generation.

Chain-of-thought reasoning with reinforcement learning reduces false alerts in security operations center detection triage.

LeanCSP formalizes constraint reformulation verification and solver correctness in the Lean theorem prover.

Self-Verifying Refinement learns adaptive test-time compute allocation via self-verification and confidence-based RL without external verifiers.

Graph neural networks improve preconditioning for sparse linear system solvers, particularly on indefinite and nonsymmetric matrices.

Aging-Aware Autonomous Intelligence integrates hardware degradation detection into reasoning and mission planning for autonomous systems.

Lightning OPD 2.0 mitigates teacher inconsistency in on-policy distillation for reasoning models via cross-teacher style bias correction.

Evidence-Grounded Social Persona Panel evaluates generative UI quality across psychologically diverse user representations.

Short-term graph memory module optimizes molecular search by pre-screening candidates under fixed oracle budget constraints.

Analysis of LLM hidden states using aggregator/differentiator metrics to measure how tokens consolidate text representation and metaphorical transport across positions.

UNICON foundation model demonstrates numerical intelligence via in-context learning across scientific/social datasets, generalizing beyond language-only reasoning.

The deal gives Okta identity threat detection capabilities as enterprises seek to secure AI agents and other non-human identities across cloud environments.

Physics-inspired KSSE method replaces dense CNN classifiers using sparse-graph spectral embedding and Ising models for image classification.

READII-2-ROQC framework uses negative controls to detect volume confounding in radiomics and imaging foundation model biomarkers.

QAdapt neural pre-decoding framework adapts to nonstationary hardware noise in quantum error correction via noise-adaptive decoders.

Study of derived-feature over-trust (DFOT) in LLMs using physiological sensing; proposes privileged-modality reliability evidence to mitigate misapplication.

WIDE framework enables token-level dynamic width pruning for adaptive LLM inference, improving efficiency while preserving accuracy under aggressive sparsity.

QQWorld replaces Epps-Pulley objective with quantile-quantile matching to regularize latent world model distributions for improved planning.

Two AI computing clusters built from identical NVIDIA H100, GB200 NVL72, or GB300 NVL72 systems can deliver materially different training throughput. We... Two AI computing clusters built from identical NVIDIA H100, GB200 NVL72, or GB300 NVL72 systems can deliver materially different training throughput. We routinely see 8% to 12% gaps between partner deployments and the corresponding NVIDIA reference architecture (RA) on the same workload, same model, same global batch size. The cause is often a stack of configuration choices in the kernel… Source

Query complexity analysis of windowed thinning for bouncy particle and Zigzag samplers under strongly convex/smooth potentials.

PACE hierarchical framework applies LLMs to parent-order execution in algorithmic trading, generalizing across market conditions without task-specific training.