Kenji Kawaguchi

Affiliations

Massachusetts Institute of Technology

papers

886

total citations

papers (29)

Self-Evaluation Guided Beam Search for Reasoning

NEURIPS 2023arXiv

244

citations

Knowledge-Augmented Reasoning Distillation for Small Language Models in Knowledge-Intensive Tasks

NEURIPS 2023arXiv

citations

Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety Tuning

ICLR 2025arXiv

citations

MGNNI: Multiscale Graph Neural Networks with Implicit Layers

NEURIPS 2022arXiv

citations

Noether Networks: meta-learning useful conserved quantities

NEURIPS 2021arXiv

citations

Understanding and Enhancing Safety Mechanisms of LLMs via Safety-Specific Neuron

ICLR 2025

citations

Drug Discovery with Dynamic Goal-aware Fragments

ICML 2024arXiv

citations

Enhancing Semantic Fidelity in Text-to-Image Synthesis: Attention Regulation in Diffusion Models

ECCV 2024arXiv

citations

The Surprising Effectiveness of Skip-Tuning in Diffusion Sampling

ICML 2024arXiv

citations

Set-based Meta-Interpolation for Few-Task Meta-Learning

NEURIPS 2022arXiv

citations

Unsupervised Concept Discovery Mitigates Spurious Correlations

ICML 2024arXiv

citations

VA3: Virtually Assured Amplification Attack on Probabilistic Copyright Protection for Text-to-Image Generative Models

CVPR 2024arXiv

citations

Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding

AAAI 2024arXiv

citations

Exact Conversion of In-Context Learning to Model Weights in Linearized-Attention Transformers

ICML 2024arXiv

citations

PlanDQ: Hierarchical Plan Orchestration via D-Conductor and Q-Performer

ICML 2024arXiv

citations

Referee Can Play: An Alternative Approach to Conditional Generation via Model Inversion

ICML 2024arXiv

citations

Unnatural Languages Are Not Bugs but Features for LLMs

ICML 2025arXiv

citations

Deep Regression Representation Learning with Topology

ICML 2024arXiv

citations

PICProp: Physics-Informed Confidence Propagation for Uncertainty Quantification

NEURIPS 2023arXiv

citations

An Information Theory Perspective on Variance-Invariance-Covariance Regularization

NEURIPS 2023

citations

The Stronger the Diffusion Model, the Easier the Backdoor: Data Poisoning to Induce Copyright BreachesWithout Adjusting Finetuning Pipeline

ICML 2024

citations

Discrete Compositional Representations as an Abstraction for Goal Conditioned Reinforcement Learning

NEURIPS 2022

citations

Understanding End-to-End Model-Based Reinforcement Learning Methods as Implicit Parameterization

NEURIPS 2021

citations

Tailoring: encoding inductive biases by optimizing unsupervised objectives at prediction time

NEURIPS 2021arXiv

citations

Kenji Kawaguchi

Affiliations

papers (29)

Self-Evaluation Guided Beam Search for Reasoning

Knowledge-Augmented Reasoning Distillation for Small Language Models in Knowledge-Intensive Tasks

Towards 3D Molecule-Text Interpretation in Language Models

Adversarial Training Helps Transfer Learning via Better Representations

Rethinking Tokenizer and Decoder in Masked Graph Modeling for Molecules

Discrete-Valued Neural Communication

EIGNN: Efficient Infinite-Depth Graph Neural Networks

Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety Tuning

MGNNI: Multiscale Graph Neural Networks with Implicit Layers

Noether Networks: meta-learning useful conserved quantities

Understanding and Enhancing Safety Mechanisms of LLMs via Safety-Specific Neuron

Drug Discovery with Dynamic Goal-aware Fragments

Enhancing Semantic Fidelity in Text-to-Image Synthesis: Attention Regulation in Diffusion Models

The Surprising Effectiveness of Skip-Tuning in Diffusion Sampling

Set-based Meta-Interpolation for Few-Task Meta-Learning

Unsupervised Concept Discovery Mitigates Spurious Correlations

VA3: Virtually Assured Amplification Attack on Probabilistic Copyright Protection for Text-to-Image Generative Models

Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding

Exact Conversion of In-Context Learning to Model Weights in Linearized-Attention Transformers

PlanDQ: Hierarchical Plan Orchestration via D-Conductor and Q-Performer

Referee Can Play: An Alternative Approach to Conditional Generation via Model Inversion

Unnatural Languages Are Not Bugs but Features for LLMs

Deep Regression Representation Learning with Topology

PICProp: Physics-Informed Confidence Propagation for Uncertainty Quantification

An Information Theory Perspective on Variance-Invariance-Covariance Regularization

The Stronger the Diffusion Model, the Easier the Backdoor: Data Poisoning to Induce Copyright BreachesWithout Adjusting Finetuning Pipeline

Discrete Compositional Representations as an Abstraction for Goal Conditioned Reinforcement Learning

Understanding End-to-End Model-Based Reinforcement Learning Methods as Implicit Parameterization

Tailoring: encoding inductive biases by optimizing unsupervised objectives at prediction time

papers (29)

Self-Evaluation Guided Beam Search for Reasoning

Knowledge-Augmented Reasoning Distillation for Small Language Models in Knowledge-Intensive Tasks

Towards 3D Molecule-Text Interpretation in Language Models

Adversarial Training Helps Transfer Learning via Better Representations

Rethinking Tokenizer and Decoder in Masked Graph Modeling for Molecules

Discrete-Valued Neural Communication

EIGNN: Efficient Infinite-Depth Graph Neural Networks

Learning Diverse Attacks on Large Language Models for Robust Red-Teaming and Safety Tuning

MGNNI: Multiscale Graph Neural Networks with Implicit Layers

Noether Networks: meta-learning useful conserved quantities

Understanding and Enhancing Safety Mechanisms of LLMs via Safety-Specific Neuron

Drug Discovery with Dynamic Goal-aware Fragments

Enhancing Semantic Fidelity in Text-to-Image Synthesis: Attention Regulation in Diffusion Models

The Surprising Effectiveness of Skip-Tuning in Diffusion Sampling

Set-based Meta-Interpolation for Few-Task Meta-Learning

Unsupervised Concept Discovery Mitigates Spurious Correlations

VA3: Virtually Assured Amplification Attack on Probabilistic Copyright Protection for Text-to-Image Generative Models

Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding

Exact Conversion of In-Context Learning to Model Weights in Linearized-Attention Transformers

PlanDQ: Hierarchical Plan Orchestration via D-Conductor and Q-Performer

Referee Can Play: An Alternative Approach to Conditional Generation via Model Inversion

Unnatural Languages Are Not Bugs but Features for LLMs

Deep Regression Representation Learning with Topology

PICProp: Physics-Informed Confidence Propagation for Uncertainty Quantification

An Information Theory Perspective on Variance-Invariance-Covariance Regularization

The Stronger the Diffusion Model, the Easier the Backdoor: Data Poisoning to Induce Copyright BreachesWithout Adjusting Finetuning Pipeline

Discrete Compositional Representations as an Abstraction for Goal Conditioned Reinforcement Learning

Understanding End-to-End Model-Based Reinforcement Learning Methods as Implicit Parameterization

Tailoring: encoding inductive biases by optimizing unsupervised objectives at prediction time