PARADIGM_SHIFT

329 papers · Page 2 of 4

Unifies large-scale search, recommendation, and reasoning into a single self-contained LLM by treating item IDs as a distinct modality.

AI & ML arxiv | Mar 19

Edit-As-Act reframes 3D scene editing as a goal-regressive planning problem using symbolic action languages rather than purely generative pixel manipulation.

AI & ML arxiv | Mar 19

A new self-refining surrogate framework enables neural models to simulate complex dynamical systems over arbitrarily long horizons without the standard failure mode of compounding error.

AI & ML arxiv | Mar 19

The 'consensus trap' in label-free RL—where models reinforce their own systematic errors—can be broken by co-evolving the model in alternating generator and verifier roles.

AI & ML arxiv | Mar 19

LLMs compute and cache confidence scores automatically during answer generation, well before they are prompted to verbalize them.

AI & ML arxiv | Mar 19

Measuring the distance between human languages can now be done quantitatively using the attention mechanisms of multilingual transformers.

AI & ML arxiv | Mar 19

AgentFactory shifts agent evolution from unreliable textual 'reflections' to a library of verifiable, executable Python subagents.

AI & ML arxiv | Mar 19

DAPS++ reinterprets diffusion inverse problems as a decoupled EM-style initialization, significantly increasing restoration speed and stability.

AI & ML arxiv | Mar 19

A geometric fix for Rotary Positional Embeddings (RoPE) allows Transformers to generalize to long inputs out-of-the-box by preserving 'sink token' functionality.

AI & ML arxiv | Mar 20

A synthesizable RTL implementation of Predictive Coding allows for fully distributed, non-backprop learning directly in hardware.

AI & ML arxiv | Mar 20

Dynamic constraints using an 'online refiner' resolve the conflict between stability and performance in Reinforcement Learning Fine-Tuning (RFT).

AI & ML arxiv | Mar 20

Uses Pearl's do-operator to automatically discover and mask irrelevant state dimensions in Reinforcement Learning.

AI & ML arxiv | Mar 20

Fine-tunes Vision-Language Models using raw images alone by using a text-to-image model as a cycle-consistency reward.

AI & ML arxiv | Mar 20

PowerFlow uses GFlowNets to replace heuristic rewards in unsupervised fine-tuning, allowing practitioners to explicitly tune models for either logic or creativity.

AI & ML arxiv | Mar 20

AS2 achieves a fully differentiable neuro-symbolic bridge by replacing discrete solvers with a soft, continuous approximation of the Answer Set Programming operator.

AI & ML arxiv | Mar 20

Standard decoding strategies (top-k, nucleus) create a 'truncation blind spot' by systematically excluding human-like, low-probability token choices.

AI & ML arxiv | Mar 20

SINDy-KANs combine Kolmogorov-Arnold Networks with Sparse Identification of Non-linear Dynamics to create parsimonious, interpretable models.

AI & ML arxiv | Mar 20

REST transforms the zero-shot object-navigation problem from simple waypoint selection to a tree-of-paths reasoning process.

AI & ML arxiv | Mar 20

A linear-time attention mechanism that is weight-compatible with standard pretrained Transformers, allowing for direct knowledge transfer.

AI & ML arxiv | Mar 20

A system where agents autonomously design, refine, and store task-specific skills as 'stateful prompts' to achieve non-parametric continual learning.

AI & ML arxiv | Mar 20

Shifts concept unlearning in diffusion models from fragile keyword-based removal to a distributional framework using contextually diverse prompts.

AI & ML arxiv | Mar 20

Eliminates the need for expensive process reward models by propagating terminal rewards across state-space graphs to generate dense, state-level rewards for agentic RL.

AI & ML arxiv | Mar 20

Introduces 'intentional interventions' and Structural Final Models (SFMs) to detect and infer agent goals within causal frameworks.

AI & ML arxiv | Mar 20

Uses Sparse Autoencoders (SAEs) to disentangle and modulate bias-relevant features in Vision-Language Models without retraining.

AI & ML arxiv | Mar 20

Incorporates the physics of forward dynamics directly into a GNN architecture for articulated robot control.

AI & ML arxiv | Mar 20

Argues that standard ML efficiency metrics (FLOPs, throughput) are poorly correlated with actual robot performance in Vision-Language-Action (VLA) models.

AI & ML arxiv | Mar 20

Reframes GPU kernel optimization by benchmarking against hardware 'Speed-of-Light' limits rather than software baselines.

AI & ML arxiv | Mar 20

Repurposes pre-trained video diffusion models as 'Latent World Simulators' to give Multimodal LLMs 3D spatial awareness without explicit 3D data.

AI & ML arxiv | Mar 20

Introduces a statistical alternative to the standard frequency-based BPE tokenization used in nearly all modern LLMs.

AI & ML arxiv | Mar 23

Formally proves that a causal Transformer is mathematically equivalent to a stateless Differentiable Neural Computer.

AI & ML arxiv | Mar 23

Solves the compositional generalization failure of neural networks (0% to 100% accuracy) by embedding algebraic semiring constraints.

AI & ML arxiv | Mar 23

Challenges the 80-year-old assumption that neurons must use weighted summation as their primary aggregation mechanism.

AI & ML arxiv | Mar 23

Introduces Hyperagents: self-referential systems where the meta-level modification logic is itself an editable program.

AI & ML arxiv | Mar 23

Fine-tunes Large Vision Language Models for medical tasks using only image-description pairs, bypassing the need for expensive expert-curated instructions.

AI & ML arxiv | Mar 23

Formalizes the 'Neural Uncertainty Principle,' linking adversarial vulnerability in vision and hallucinations in LLMs to a shared geometric and information-theoretic origin.

AI & ML arxiv | Mar 23

A massive field study (9,000+ users) proves that algorithmic shifts can reduce affective polarization without sacrificing user engagement.

AI & ML arxiv | Mar 23

Enables zero-shot humanoid robot interaction by generating robot-centric 'dream' videos instead of relying on human-to-robot motion retargeting.

AI & ML arxiv | Mar 23

Replaces fixed context compression ratios with a performance-floor constraint to ensure reliable LLM deployment.

AI & ML arxiv | Mar 23

FIPO overcomes reasoning length stagnation in LLMs by using Future-KL divergence to create dense rewards, extending Chain-of-Thought lengths to over 10,000 tokens.

AI & ML arxiv | Mar 23

Breaking the 'capability ceiling' in LLM post-training by replacing full-history dependencies with explicit Markov states.

AI & ML arxiv | Mar 23

Identifies 'critical times' in diffusion generation where targeted guidance pulses significantly improve image control.

AI & ML arxiv | Mar 23

Derives a variational ELBO for the Joint-Embedding Predictive Architecture (JEPA), unifying it with generative modeling.

AI & ML arxiv | Mar 23

Integrates Kolmogorov-Arnold Networks (KANs) into causal generative modeling to produce human-readable symbolic structural equations.

AI & ML arxiv | Mar 23

Introduces a training strategy where Transformers 'think' in latent space before committing to discrete tokens.

AI & ML arxiv | Mar 24

The first foundation model for zero-shot prediction of joint probability distributions in coupled time series.

AI & ML arxiv | Mar 24

Formalizes 'Introspection' in LLMs and proves they have privileged access to their own policy logic beyond mere self-simulation.

AI & ML arxiv | Mar 24

Reason-to-Transmit introduces deliberative communication for multi-agent systems, where agents reason about why a message benefits the receiver rather than just broadcasting features.

AI & ML arxiv | Mar 24

This paper demonstrates that Model Context Protocol (MCP) can outperform traditional RAG for quantitative financial Q&A by interacting directly with structured data APIs.

AI & ML arxiv | Mar 24

Leum-VL-8B introduces a structural 'grammar' for video parsing by decomposing content into six film-production-style dimensions like camera language and editing.

AI & ML arxiv | Mar 24

This paper shows that pretrained monocular models can perform multi-view human mesh recovery without camera calibration or multi-view training data.

AI & ML arxiv | Mar 24

Latent representations of reasoning survive cross-architecture translation, allowing student models to inherit teacher capabilities without training.

AI & ML arxiv | Mar 24

Coding agents navigating a file system outperform SOTA long-context LLMs and RAG systems on massive datasets.

AI & ML arxiv | Mar 24

Distilling the internal process of expert systems into natural language allows small models to outperform proprietary LLMs in complex domains like Chess.

AI & ML arxiv | Mar 24

ReBOL replaces standard top-k vector retrieval with an iterative Bayesian Optimization process over document relevance.

AI & ML arxiv | Mar 24

Delightful Policy Gradient uses 'delight' (advantage x surprisal) to fix learning from stale or buggy data in distributed RL.

AI & ML arxiv | Mar 24

Continued Fraction Neural Networks (CFNN) introduce a rational inductive bias that handles singularities with 10-100x fewer parameters than standard MLPs.

AI & ML arxiv | Mar 24

Network-of-Thought (NoT) moves LLM reasoning from linear chains and trees to complex directed graphs, significantly improving multi-hop QA.

AI & ML arxiv | Mar 24

Proposes 'semantic sections' as a replacement for global feature vectors to interpret LLMs in complex, non-linear representation spaces.

AI & ML arxiv | Mar 24

Introduces Bayesian scattering as a mathematically grounded, non-learned baseline for image uncertainty quantification.

AI & ML arxiv | Mar 24

A red-teaming protocol that uses RL-driven 'profit' objectives to find structural exploits in AI agents instead of just prompt-injection vulnerabilities.

AI & ML arxiv | Mar 24

Pretrained Diffusion Transformers (DiTs) possess an intrinsic 'synchronization gap' where different features commit at specific, depth-localized layers.

AI & ML arxiv | Mar 24

The 'routing paradox' proves that selective attention requires the very pairwise computations it aims to replace, explaining why pure recurrent models fail at associative recall.

AI & ML arxiv | Mar 24

VAE tokenizers in Latent Diffusion Models create 'overly compact' manifolds that cause variance collapse, leading to unstable generative sampling.

AI & ML arxiv | Mar 24

CounterScene endows generative world models with explicit counterfactual reasoning for safety-critical driving evaluation.

AI & ML arxiv | Mar 24

Proposes multi-cluster memory for test-time adaptation, proving that a single unstructured memory pool is fundamentally insufficient for non-i.i.d. data streams.

AI & ML arxiv | Mar 24

Reframes plasticity loss in Reinforcement Learning as an optimization problem where networks get trapped in local optima of previous tasks.

AI & ML arxiv | Mar 24

Repurposes a 2B-parameter latent video transformer as a differentiable physics simulator for urban wind flow optimization.

AI & ML arxiv | Mar 24

Proposes replacing flat conversation histories with a tree-based architecture to solve 'logical context poisoning.'

AI & ML arxiv | Mar 24

Replaces self-attention with Reaction-Diffusion PDEs as the predictive engine for world models.

AI & ML arxiv | Mar 24

Reconceptualizes human-agent interaction as dynamically generated software rather than just chat.

AI & ML arxiv | Mar 24

ADARUBRIC generates task-specific evaluation rubrics on the fly, significantly outperforming static rubrics in human correlation and agent training outcomes.

AI & ML arxiv | Mar 24

DSPA performs preference alignment at inference time by steering Sparse Autoencoder (SAE) features, bypassing the need for expensive weight-update training.

AI & ML arxiv | Mar 24

Introduces a counterfactual framework for precise individual credit assignment in collaborative multi-agent LLM systems.

AI & ML arxiv | Mar 24

Provides the first unified theoretical formalism for hierarchical memory systems used by long-context language agents.

AI & ML arxiv | Mar 24

Rule-State Inference (RSI) inverts the standard ML paradigm by treating known regulatory rules as priors and inferring the latent state of compliance and drift, rather than approximating rules from noisy data.

AI & ML arxiv | Mar 24

GSB-PPO lifts proximal policy optimization from discrete action steps to full generation trajectories by framing it as a Generalized Schrödinger Bridge.

AI & ML arxiv | Mar 24

PRM-as-a-Judge shifts robotic evaluation from binary success/failure to a dense, potential-based progress metric system.

AI & ML arxiv | Mar 24

FIM-Merging provides a theoretical framework for layer-adaptive model merging using the Fisher Information Matrix to bound merging error.

AI & ML arxiv | Mar 24

Hypothesizes and demonstrates a unified Gaussian latent geometry connecting vision encoders and generative models.

AI & ML arxiv | Mar 24

Solves the structural redundancy problem in symbolic regression by collapsing expression DAG isomorphisms.

AI & ML arxiv | Mar 24

Synergizes prompt optimization with policy optimization to overcome the 'sparse reward' problem in complex reasoning tasks.

AI & ML arxiv | Mar 24

Identifies the 'golden subspace' for test-time adaptation, enabling extreme efficiency in online model updates.

AI & ML arxiv | Mar 24

Decouples high-level reasoning from low-level motor control in robotics using a visual prompting interface.

AI & ML arxiv | Mar 24

Proposed a test-time scaling paradigm for image restoration that allows compute-to-quality trade-offs during inference.

AI & ML arxiv | Mar 24

Identifies that the direction of log-probability change is more critical than magnitude for improving LLM reasoning via RL.

AI & ML arxiv | Mar 24

Identifies 'Visual Anchor Collapse' in DPO-aligned VLMs and introduces an asymmetric constraint to prevent models from ignoring visual evidence in favor of language priors.

AI & ML arxiv | Mar 24

Bypasses Reinforcement Learning during the exploration phase by using uncertainty-guided tree search to discover informative data.

AI & ML arxiv | Mar 24

UNITE enables single-stage joint training of the tokenizer and the diffusion model from scratch, removing the need for frozen VAEs.

AI & ML arxiv | Mar 24

LassoFlexNet matches or beats leading tree-based models on tabular data while maintaining Lasso-like interpretability through per-feature embeddings and a group Lasso mechanism.

AI & ML arxiv | Mar 24

Mechanistic interpretability reveals that LLMs possess 'affect reception' circuits that detect emotional content even when explicit keywords are removed.

AI & ML arxiv | Mar 25

Gradient boosting exhibits a 'first-mover bias' where correlated features selected early in the tree sequence gain an artificial, self-reinforcing importance in SHAP rankings.

AI & ML arxiv | Mar 25

Establishes a formal mathematical equivalence between Classifier-Free Guidance (CFG) and alignment-based objectives, allowing for CFG-like quality without inference-time overhead.

AI & ML arxiv | Mar 25

Shifts symbolic regression from discrete genetic search to a continuous, embedding-driven optimization paradigm.

AI & ML arxiv | Mar 25

Replaces standard autoregressive document OCR with a parallel diffusion-based denoising framework.

AI & ML arxiv | Mar 25

Demonstrates that Hebbian plasticity can induce emergent attractor dynamics in robot controllers, enabling rapid adaptation without backpropagation.

AI & ML arxiv | Mar 25

Instead of using top-activating examples, this method steers Sparse Autoencoder (SAE) features in Vision-Language Models to let the model describe its own internal visual features.

AI & ML arxiv | Mar 25

DeIllusionLLM introduces task-level autoregressive reasoning to prevent LLMs from hallucinating answers to ill-posed or faulty scientific questions.

AI & ML arxiv | Mar 25

Inter-Layer Structural Encoders (ILSE) use Cayley graphs to aggregate features from all internal LLM layers, improving accuracy by up to 44% over final-layer-only predictions.

AI & ML arxiv | Mar 25

Introduces Dual Q-DM, the first non-adversarial imitation learning method theoretically guaranteed to eliminate compounding errors.

AI & ML arxiv | Mar 25

Moving beyond coarse reward signals, this paper introduces token-level policy optimization for multimodal reasoning.

AI & ML arxiv | Mar 25

PARADIGM_SHIFT

Unifies large-scale search, recommendation, and reasoning into a single self-contained LLM by treating item IDs as a distinct modality.

Edit-As-Act reframes 3D scene editing as a goal-regressive planning problem using symbolic action languages rather than purely generative pixel manipulation.

A new self-refining surrogate framework enables neural models to simulate complex dynamical systems over arbitrarily long horizons without the standard failure mode of compounding error.

The 'consensus trap' in label-free RL—where models reinforce their own systematic errors—can be broken by co-evolving the model in alternating generator and verifier roles.

LLMs compute and cache confidence scores automatically during answer generation, well before they are prompted to verbalize them.

Measuring the distance between human languages can now be done quantitatively using the attention mechanisms of multilingual transformers.

AgentFactory shifts agent evolution from unreliable textual 'reflections' to a library of verifiable, executable Python subagents.

DAPS++ reinterprets diffusion inverse problems as a decoupled EM-style initialization, significantly increasing restoration speed and stability.

A geometric fix for Rotary Positional Embeddings (RoPE) allows Transformers to generalize to long inputs out-of-the-box by preserving 'sink token' functionality.

A synthesizable RTL implementation of Predictive Coding allows for fully distributed, non-backprop learning directly in hardware.

Dynamic constraints using an 'online refiner' resolve the conflict between stability and performance in Reinforcement Learning Fine-Tuning (RFT).

Uses Pearl's do-operator to automatically discover and mask irrelevant state dimensions in Reinforcement Learning.

Fine-tunes Vision-Language Models using raw images alone by using a text-to-image model as a cycle-consistency reward.

PowerFlow uses GFlowNets to replace heuristic rewards in unsupervised fine-tuning, allowing practitioners to explicitly tune models for either logic or creativity.

AS2 achieves a fully differentiable neuro-symbolic bridge by replacing discrete solvers with a soft, continuous approximation of the Answer Set Programming operator.

Standard decoding strategies (top-k, nucleus) create a 'truncation blind spot' by systematically excluding human-like, low-probability token choices.

SINDy-KANs combine Kolmogorov-Arnold Networks with Sparse Identification of Non-linear Dynamics to create parsimonious, interpretable models.

REST transforms the zero-shot object-navigation problem from simple waypoint selection to a tree-of-paths reasoning process.

A linear-time attention mechanism that is weight-compatible with standard pretrained Transformers, allowing for direct knowledge transfer.

A system where agents autonomously design, refine, and store task-specific skills as 'stateful prompts' to achieve non-parametric continual learning.

Shifts concept unlearning in diffusion models from fragile keyword-based removal to a distributional framework using contextually diverse prompts.

Eliminates the need for expensive process reward models by propagating terminal rewards across state-space graphs to generate dense, state-level rewards for agentic RL.

Introduces 'intentional interventions' and Structural Final Models (SFMs) to detect and infer agent goals within causal frameworks.

Uses Sparse Autoencoders (SAEs) to disentangle and modulate bias-relevant features in Vision-Language Models without retraining.

Incorporates the physics of forward dynamics directly into a GNN architecture for articulated robot control.

Argues that standard ML efficiency metrics (FLOPs, throughput) are poorly correlated with actual robot performance in Vision-Language-Action (VLA) models.

Reframes GPU kernel optimization by benchmarking against hardware 'Speed-of-Light' limits rather than software baselines.

Repurposes pre-trained video diffusion models as 'Latent World Simulators' to give Multimodal LLMs 3D spatial awareness without explicit 3D data.

Introduces a statistical alternative to the standard frequency-based BPE tokenization used in nearly all modern LLMs.

Formally proves that a causal Transformer is mathematically equivalent to a stateless Differentiable Neural Computer.

Solves the compositional generalization failure of neural networks (0% to 100% accuracy) by embedding algebraic semiring constraints.

Challenges the 80-year-old assumption that neurons must use weighted summation as their primary aggregation mechanism.

Introduces Hyperagents: self-referential systems where the meta-level modification logic is itself an editable program.

Fine-tunes Large Vision Language Models for medical tasks using only image-description pairs, bypassing the need for expensive expert-curated instructions.

Formalizes the 'Neural Uncertainty Principle,' linking adversarial vulnerability in vision and hallucinations in LLMs to a shared geometric and information-theoretic origin.

A massive field study (9,000+ users) proves that algorithmic shifts can reduce affective polarization without sacrificing user engagement.

Enables zero-shot humanoid robot interaction by generating robot-centric 'dream' videos instead of relying on human-to-robot motion retargeting.

Replaces fixed context compression ratios with a performance-floor constraint to ensure reliable LLM deployment.

FIPO overcomes reasoning length stagnation in LLMs by using Future-KL divergence to create dense rewards, extending Chain-of-Thought lengths to over 10,000 tokens.

Breaking the 'capability ceiling' in LLM post-training by replacing full-history dependencies with explicit Markov states.

Identifies 'critical times' in diffusion generation where targeted guidance pulses significantly improve image control.

Derives a variational ELBO for the Joint-Embedding Predictive Architecture (JEPA), unifying it with generative modeling.

Integrates Kolmogorov-Arnold Networks (KANs) into causal generative modeling to produce human-readable symbolic structural equations.

Introduces a training strategy where Transformers 'think' in latent space before committing to discrete tokens.

The first foundation model for zero-shot prediction of joint probability distributions in coupled time series.

Formalizes 'Introspection' in LLMs and proves they have privileged access to their own policy logic beyond mere self-simulation.

Reason-to-Transmit introduces deliberative communication for multi-agent systems, where agents reason about *why* a message benefits the receiver rather than just broadcasting features.

This paper demonstrates that Model Context Protocol (MCP) can outperform traditional RAG for quantitative financial Q&A by interacting directly with structured data APIs.

Leum-VL-8B introduces a structural 'grammar' for video parsing by decomposing content into six film-production-style dimensions like camera language and editing.

This paper shows that pretrained monocular models can perform multi-view human mesh recovery without camera calibration or multi-view training data.

Latent representations of reasoning survive cross-architecture translation, allowing student models to inherit teacher capabilities without training.

Coding agents navigating a file system outperform SOTA long-context LLMs and RAG systems on massive datasets.

Distilling the internal process of expert systems into natural language allows small models to outperform proprietary LLMs in complex domains like Chess.

ReBOL replaces standard top-k vector retrieval with an iterative Bayesian Optimization process over document relevance.

Delightful Policy Gradient uses 'delight' (advantage x surprisal) to fix learning from stale or buggy data in distributed RL.

Continued Fraction Neural Networks (CFNN) introduce a rational inductive bias that handles singularities with 10-100x fewer parameters than standard MLPs.

Network-of-Thought (NoT) moves LLM reasoning from linear chains and trees to complex directed graphs, significantly improving multi-hop QA.

Proposes 'semantic sections' as a replacement for global feature vectors to interpret LLMs in complex, non-linear representation spaces.

Introduces Bayesian scattering as a mathematically grounded, non-learned baseline for image uncertainty quantification.

A red-teaming protocol that uses RL-driven 'profit' objectives to find structural exploits in AI agents instead of just prompt-injection vulnerabilities.

Pretrained Diffusion Transformers (DiTs) possess an intrinsic 'synchronization gap' where different features commit at specific, depth-localized layers.

The 'routing paradox' proves that selective attention requires the very pairwise computations it aims to replace, explaining why pure recurrent models fail at associative recall.

VAE tokenizers in Latent Diffusion Models create 'overly compact' manifolds that cause variance collapse, leading to unstable generative sampling.

CounterScene endows generative world models with explicit counterfactual reasoning for safety-critical driving evaluation.

Proposes multi-cluster memory for test-time adaptation, proving that a single unstructured memory pool is fundamentally insufficient for non-i.i.d. data streams.

Reframes plasticity loss in Reinforcement Learning as an optimization problem where networks get trapped in local optima of previous tasks.

Repurposes a 2B-parameter latent video transformer as a differentiable physics simulator for urban wind flow optimization.

Proposes replacing flat conversation histories with a tree-based architecture to solve 'logical context poisoning.'

Replaces self-attention with Reaction-Diffusion PDEs as the predictive engine for world models.

Reconceptualizes human-agent interaction as dynamically generated software rather than just chat.

ADARUBRIC generates task-specific evaluation rubrics on the fly, significantly outperforming static rubrics in human correlation and agent training outcomes.

DSPA performs preference alignment at inference time by steering Sparse Autoencoder (SAE) features, bypassing the need for expensive weight-update training.

Introduces a counterfactual framework for precise individual credit assignment in collaborative multi-agent LLM systems.

Provides the first unified theoretical formalism for hierarchical memory systems used by long-context language agents.

Rule-State Inference (RSI) inverts the standard ML paradigm by treating known regulatory rules as priors and inferring the latent state of compliance and drift, rather than approximating rules from noisy data.

GSB-PPO lifts proximal policy optimization from discrete action steps to full generation trajectories by framing it as a Generalized Schrödinger Bridge.

PRM-as-a-Judge shifts robotic evaluation from binary success/failure to a dense, potential-based progress metric system.

FIM-Merging provides a theoretical framework for layer-adaptive model merging using the Fisher Information Matrix to bound merging error.

Hypothesizes and demonstrates a unified Gaussian latent geometry connecting vision encoders and generative models.

Reason-to-Transmit introduces deliberative communication for multi-agent systems, where agents reason about why a message benefits the receiver rather than just broadcasting features.