COGENT is a new architecture combining graph neural networks with Neural Ordinary Differential Equations for continuous-time physical forecasting on irregular geospatial meshes. The model encodes historical system states and forcings into latent dynamics that can be queried at arbitrary future times, avoiding the error accumulation of autoregressive rollout. Evaluated on ice-sheet simulations from the Ice-sheet and Sea-level System Model, COGENT shows improved long-range stability over autoregressive graph baselines. The work introduces training stabilization strategies including rollout-horizon sampling and progressive scheduling.
Cognee is an open-source Python library providing persistent long-term memory for AI agents via a self-hosted knowledge graph engine. The project has accumulated 18,495 GitHub stars with 361 added today, indicating significant community traction. It targets the agent memory problem by enabling cross-session state persistence without reliance on external managed services.
A preprint proposes a fully automated time-series prediction architecture for Cloud-Edge Continuum (CEC) orchestration, addressing the cold-start problem where newly discovered edge nodes lack historical data for localized model training. The system combines a lightweight Resource Exposer for telemetry collection with a novel data-mixing methodology that merges sparse local samples with TimeTrack, a publicly released high-resolution dataset, then feeds the result through a Neural Architecture Search engine to auto-generate baseline models. Experiments show the approach improves MSE, MAE, and MAPE and accelerates convergence versus training on local data alone or generic datasets.
PolyGnosis 2.0 introduces a multi-agent system that synthesizes Polymarket prediction market signals with GDELT OSINT streams to identify 'Perspective Mismatches' as trading signals. The paper rigorously evaluates agentic harness engineering techniques—reflection loops, tool-calling, divide-and-conquer partitioning, and chain-of-thought—in high-noise financial domains. Key empirical findings include that structural partitioning is necessary for multi-dimensional alignment, but unconstrained terminal reflection induces logical drift, and a pervasive consensus bias emerges across agent configurations. The authors identify a Pareto-optimal configuration achieving professional-grade analytical precision with minimized latency and token overhead.
Researchers introduce the Gubernaut Cognitive Controller (GCC), a model-agnostic runtime layer that monitors numeric telemetry (intensity, valence, repetition) at a meta level to regulate LLM agent behavior under sustained pressure—addressing escalation, sycophancy, and perseveration without modifying model weights. The architecture uses a Nelson–Narens monitoring–control loop where the deterministic meta level ingests zero tokens, eliminating a class of prompt-injection attack vectors by construction. Evaluation uses a pre-registered generate-once/judge-many protocol across a 4×4 matrix of four frontier models (GPT-5.5, Claude Opus 4.8, Gemini 3.5 Flash, Grok 4.3) as both generators and judges, finding the regulated arm calmer in 13 of 16 cells at p<.05 and 15 of 16 by sign. The recovery signature—arousal integrating under attack then decaying on de-escalation—replicates across all four model families, suggesting a robust mechanism rather than a judge-style artifact.
GS-Agent is an end-to-end multi-agent system that generates dynamic, physically plausible 4D worlds from natural language descriptions by integrating physics engines in the loop. The framework decomposes world creation into entity management (3D asset curation, material tuning, placement, motion control) and rendering configuration, with specialized agents collaborating via code and multimodal feedback. Results demonstrate generation of scenes with rich interactions among liquids, deformable objects, and rigid bodies, with cinematic camera and lighting control. The work positions itself as a foundation for physical AI and automated creative content generation.
Researchers propose training attention graph neural networks (GNNs) as neural surrogates for 3D coupled thermo-fluid fields by minimizing finite-volume method (FVM) residuals directly on the mesh, eliminating the need for labeled CFD simulation data. The approach is evaluated across four scenarios including steady-state and parametric transient cases, achieving 2.3–2.8% normalized RMSE on steady benchmarks and outperforming a supervised baseline on transient cases. The method reduces model development cost by avoiding expensive data generation from conventional numerical solvers.
Researchers introduce Controllable Neural Variational Agents (CNeVA), a framework for traffic simulation that infers per-agent Gaussian behavior latents from discounted returns via conjugate variational updates, conditioning a rectified-flow trajectory generator with classifier-free guidance. The system enables interpretable steering along axes such as speed, acceleration, and safety compliance without sacrificing realism. Evaluated on the Waymo Open Motion Dataset, CNeVA achieves competitive realism while offering per-channel controllability absent from higher-ranked imitation baselines. The work also introduces soft eligibility gates to address reward signal scarcity near decision thresholds.
A new arXiv preprint introduces GatedLinear, a lightweight forecasting framework that routes inputs across three specialized linear mechanisms — trend-seasonal, incremental drift, and phase-aligned recurrence — using a Tri-Factorized Fusion Gate. The gate disentangles routing decisions into channel-specific, horizon-aware, and phase-indexed components without heavy neural modules. Experiments on standard benchmarks claim state-of-the-art or competitive accuracy against larger foundational models with a substantially smaller parameter footprint and interpretable routing patterns.