Paradigm Shift

Proposes SOL-Nav, which replaces raw visual features in navigation with structured language descriptions for LLM-based agents.

Sci-Mind introduces an 'Adversarial Cognitive Dialectic' where specialized agents debate to refine mathematical models.

Introduces 'Umwelt Engineering,' the deliberate constraint of an agent's linguistic environment to improve reasoning.

Introduces Composer, a paradigm that generates input-specific parameter adaptations at inference time to enable dynamic per-input model specialization.

SkyNet extends MuZero to partially-observable stochastic games by adding auxiliary belief-aware heads, significantly outperforming baselines in complex card games.

The Physics-Guided Transformer (PGT) embeds physical priors (like diffusion and causality) directly into the self-attention mechanism via heat-kernel biases.

SARL improves reasoning models by rewarding the 'topology' of thoughts rather than just the final answer, enabling effective RL without ground-truth labels.

Correlated Diffusion replaces independent noise with structured MCMC dynamics, enabling generative modeling on hyper-efficient probabilistic computers.

This paper clarifies that Diffusion Maps (DMAPs) are not actually a dimensionality reduction tool, but rather a spectral representation that requires specific combinations to form a chart.

PhysNet embeds physical tumor growth dynamics directly into the latent feature space of a CNN, rather than just as a constraint on the output.

This paper proves that reward hacking is a structural equilibrium of optimized AI agents, not a bug, and provides a computable 'distortion index' to predict it.

Moves VLM grounding from text-based coordinates to a direct visual token selection mechanism via special pointing tokens.

Bypasses expensive formal verification solvers by designing neural networks that are 'verifiable by design' using the fast trivial Lipschitz bound.

Replaces traditional fixed-update rules in online learning with a causal Transformer to track switching experts in non-stationary environments.

Moves beyond next-token prediction to model reasoning as gradient-based energy minimization over latent trajectories.

Entropic Claim Resolution (ECR) shifts RAG from retrieving 'relevant' documents to retrieving 'discriminative' evidence that minimizes hypothesis uncertainty.

The 'Bidirectional Coherence Paradox' demonstrates that LLM performance and explanation quality can be inversely correlated depending on domain observability.

COvolve creates an automated curriculum for open-ended learning by co-evolving environments and policies as executable code through a zero-sum game.

Seen2Scene is the first flow matching model trained directly on incomplete real-world 3D scans rather than synthetic complete data.

Unrestrained Simplex Denoising treats discrete data generation as a non-Markovian process on the probability simplex.

PRCO decouples perception and reasoning in Multimodal RL through an Observer-Solver architecture.

SOLE-R1 uses Vision-Language Model chain-of-thought reasoning as the sole reward signal for zero-shot robotic reinforcement learning.

Metric Similarity Analysis (MSA) uses Riemannian geometry to compare the intrinsic geometry of neural representations.

Introduces a CNN architecture where feature maps are mathematically identical to Grad-CAM saliency maps by design, rather than post-hoc.

Shifts world model evaluation from visual fidelity to 'Simulative Reasoning,' revealing a massive gap in current AI's ability to plan.

Learns high-level symbolic state machines directly from raw pixels to guide robot control without hand-crafted priors.

Demonstrates that symbolic event primitives (like Schank's Conceptual Dependency) can be 'rediscovered' by neural networks purely through compression pressure.

Identifies specific hidden-state dimensions (H-Nodes) responsible for hallucinations and introduces a real-time defense to cancel them.

Moves industrial recommendation systems from static multi-stage pipelines to self-evolving agentic loops.

Empirically proves that AI Scientist agents can genuinely learn from physical experimental feedback via in-context learning.

Replaces standard autoregressive action generation in robot VLAs with iterative refinement via discrete flow matching.

Introduces a multi-agent CAD generation pipeline that uses programmatic geometric validation from the OpenCASCADE kernel to iteratively fix dimensional errors.

Introduces Process-Aware Policy Optimization (PAPO) to solve the chronic issue of reward hacking in process reward models (PRMs).

Demonstrates that perplexity/log-likelihood is a deceptive metric for model distillation, often masking massive drops in actual generation quality.

Shifts 3D scene generation from diffusion to a fully autoregressive paradigm using next-token prediction of 3D Gaussian primitives.

Proposes a universal denoiser that outperforms the Bayes-optimal Tweedie's formula when the noise distribution is unknown.

Provides the first formal proof and verification framework for agent-tool integration protocols.

Demonstrates that visual hierarchies require Lorentzian causal structure rather than Euclidean space.

Proves that Transformers can internalize complex search algorithms like MCTS directly into their weights.

Introduces a multi-answer RL objective that trains models to represent a distribution of valid answers in a single forward pass.

The 'Reasoning Contamination Effect' shows that Chain-of-Thought (CoT) reasoning actually disrupts a model's internal confidence signal, leading to poorer calibration.

R1Sim applies the 'Reasoning-RL' paradigm (popularized by DeepSeek-R1) to traffic simulation, achieving superior safety and diversity in multi-agent behaviors.

SIGMA resolves 'trajectory divergence' in molecular string generation by enforcing geometric symmetry recognition through contrastive learning.

Pixelis shifts VLM reasoning from static description to a 'reasoning in pixels' agentic paradigm that learns via an executable tool grammar.

The AE4E paradigm proposes a 'Social Contract' for multi-agent economies, replacing individual model alignment with an institutional 'Separation of Power'.

Using Signal Detection Theory, this work proves that LLM calibration and 'metacognitive efficiency' (knowing what you know) are distinct, dissociable capacities.

Vision Hopfield Memory Networks (V-HMN) present a brain-inspired alternative to Transformers and Mamba using hierarchical associative memory mechanisms.

Representing GPS trajectories as hyperspectral images enables multi-month dense anomaly detection that was previously computationally intractable.

Fixes the inherent instability of on-policy distillation in LLMs using local support matching and top-p rollout sampling.

Enables LMMs to 'think' using compact latent visual representations rather than verbalizing everything into text.