Ax Andy Zeyi Liu, Michael Zhang, Ilana Greenberg, Adam Alnasser, Lucas Baker, John Sous 5/8/2026

Memory Inception: Latent-Space KV Cache Manipulation for Steering LLMs

Training-free method for steering LLMs via latent attention space manipulation. Enables structured memory insertion without prompting overhead.

Ax Hyunjun Na, Donghwan Lee 5/8/2026

Soft Deterministic Policy Gradient with Gaussian Smoothing

Soft deterministic policy gradient method using Gaussian smoothing to handle sparse/discrete rewards in continuous control.

Ax Guanmeng Xian, Ning Yang, Philip S. Yu 5/8/2026

Band Together: Untargeted Adversarial Training with Multimodal Coordination against Evasion-based Promotion Attacks

Defense against evasion attacks in multimodal recommender systems using adversarial training with cross-modal coordination.

Ax Masatsugu Yamada, Mahito Sugiyama 5/8/2026

When Graph Language Models Go Beyond Memorization

Diagnostic framework to distinguish memorization from structural learning in graph language models using subgraph mining.

Ax Amirhossein Yousefiramandi 5/8/2026

Cumulative-Goodness Free-Riding in Forward-Forward Networks: Real, Repairable, but Not Accuracy-Dominant

Analysis of free-riding behavior in Forward-Forward neural network training. Identifies layer-wise gradient decay and proposes solutions.

Ax Jan-Hendrik Ewering, Kathrin Fla{\ss}kamp, Niklas Wahlstr\"om, Thomas B. Sch\"on, Thomas Seel 5/8/2026

Structure-Preserving Gaussian Processes Via Discrete Euler-Lagrange Equations

Research on Lagrangian Gaussian Processes for learning dynamical systems while preserving geometric structure. Original arXiv paper.

Ax Jan von Pichowski, Al\v{z}beta Hrabo\v{s}ov\'a, Ingo Scholtes, Christopher Bl\"ocker 5/8/2026

The Role of Node Features in Graph Pooling

Study analyzing how node features and graph topology interact in graph pooling for classification tasks, examining conditions for effective pooling operators.

Ax Taehun Cha, Daniel Beaglehole, Adityanarayanan Radhakrishnan, Donghun Lee 5/8/2026

The Weight Gram Matrix Captures Sequential Feature Linearization in Deep Networks

ArXiv paper proposing feature-centric framework showing weight Gram matrix captures sequential feature linearization in deep network training.

Ax Wentao Yu, Bo Han, Jie Yang, Chen Gong 5/8/2026

Beyond Rigid Alignment: Graph Federated Learning via Dual Manifold Calibration

ArXiv paper proposing dual manifold calibration for graph federated learning to handle semantic and structural heterogeneity across distributed clients.

Ax Eugenio Lomurno, Filippo Balzarini, Francesco Benelle, Francesca Pia Panaccione, Matteo Matteucci 5/8/2026

Inference-Time Refinement Closes the Synthetic-Real Gap in Tabular Diffusion

ArXiv paper on inference-time refinement for tabular diffusion models to close synthetic-real gap without retraining the generator backbone.

Ax Tyler Ingebrand, Ruihan Zhao, Kushagra Gupta, David Fridovich-Keil, Sandeep P. Chinchali, Ufuk Topcu 5/8/2026

A Flow Matching Algorithm for Many-Shot Adaptation to Unseen Distributions

ArXiv paper presenting Function Projection for Flow Matching algorithm that conditions generative models on target distribution samples for adaptation.

Ax April Chan, Davide D'Ascenzo, Sebastiano Cultrera di Montesano 5/8/2026

When Labels Have Structure: Improving Image Classification with Hierarchy-Aware Cross-Entropy

ArXiv paper proposing Hierarchy-Aware Cross-Entropy loss that incorporates class hierarchy to penalize semantically distant misclassifications differently.

Ax Jean-Loup Dupret, Davide Gallon, Patrick Cheridito 5/8/2026

INEUS: Iterative Neural Solver for High-Dimensional PIDEs

ArXiv paper presenting INEUS, a meshfree iterative neural solver for partial integro-differential equations using single-jump sampling and recursive regression.

Ax Hubert Baniecki, Przemyslaw Biecek, Fabian Fumagalli 5/8/2026

Attributions All the Way Down? The Metagame of Interpretability

ArXiv paper introducing meta-attributions framework to measure second-order interaction effects of model explanations using Shapley value game theory.

Ax Yi Wei, Xuan Qi, Furao Shen 5/8/2026

Region Seeding via Pre-Activation Regularization: A Geometric View from Piecewise Affine Nerual Networks

ArXiv paper analyzing region seeding in piecewise affine neural networks through pre-activation regularization to improve expressive capacity.

Ax Zakaria Elabid, Jan Andrzejewski, Bartosz Brzoza, Attila Cangi 5/8/2026

Molecules Meet Language: Confound-Aware Representation Learning and Chemical Property Steering in Transformer-VAE Latent Spaces

ArXiv paper on confound-aware representation learning in Transformer-VAE for molecular generation with chemical property steering in latent spaces.

Ax Siru Zhong, Zhao Meng, Haohuan Fu, Haoyang Li, Qingsong Wen, Yuxuan Liang 5/8/2026

Perceive, Route and Modulate: Dynamic Pattern Recalibration for Time Series Forecasting

ArXiv paper proposing Dynamic Pattern Recalibration for time series forecasting that adapts to shifting local temporal patterns instead of using fixed weights.

Ax Ye Su, Jian Li, Yong Liu 5/8/2026

When Does $\ell_2$-Boosting Overfit Benignly? High-Dimensional Risk Asymptotics and the $\ell_1$ Implicit Bias

ArXiv paper analyzing benign overfitting in ℓ2-boosting under ℓ1 implicit bias of greedy ensembles with high-dimensional risk asymptotics.

Ax Ruotong Sun, Ermin Wei 5/8/2026

Pro-KLShampoo: Projected KL-Shampoo with Whitening Recovered by Orthogonalization

ArXiv paper introducing Pro-KLShampoo optimizer combining Kronecker-factored preconditioning with orthogonalization for LLM pre-training efficiency.

Ax Akhil Jindal, Harang Ju 5/8/2026

SMolLM: Small Language Models Learn Small Molecular Grammar

53K-parameter transformer learning valid SMILES generation with 95% validity, outperforming larger models.

Ax Shaofeng Qin, Li Wang 5/8/2026

LINC: Decoupling Local Consequence Scoring from Hidden Matching in Constructive Neural Routing

Neural routing decoder architecture computing explicit one-step consequences for constructive solvers.

Ax Fabian Kabus, Kian Kordtomeikel, Thomas Brox, Heinz Wiendl, Daiana Stolz, Harald Binder 5/8/2026

Eliciting associations between clinical variables from LLMs via comparison questions across populations

Method to extract clinical variable associations from LLMs using structured comparison questions.

Ax Dylan Bouchard 5/8/2026

Is Escalation Worth It? A Decision-Theoretic Characterization of LLM Cascades

Decision-theoretic framework optimizing cost-quality tradeoffs in LLM cascades for model deferral.

Ax Yifan Tang, Qiquan Wang, In\'es Garc\'ia-Redondo, Anthea Monod 5/8/2026

Topological Signatures of Grokking

Topological analysis of grokking phenomenon using persistent homology on embedding matrices.

Ax Ignacio Peis, Pablo M. Olmos, Jes Frellsen 5/8/2026

Order-Agnostic Autoregressive Modelling with Missing Data

Order-agnostic autoregressive models for generative modeling with incomplete and missing data handling.

Ax Yuan Du, Mitchel Hill, HanQin Cai 5/8/2026

Memory Efficient Full-gradient Attacks (MEFA) Framework for Adversarial Defense Evaluations

Memory-efficient gradient attack framework for evaluating diffusion and Langevin-based adversarial defenses.

Ax Alessandro Pagani, Marco Cominelli, Liying Han, Gaofeng Dong, Sergio Benini, Francesco Gringoli, Mattia Savardi, Mani B. Srivastava, Trevor Bihl, Erik P. Blasch, Daniel O. Brigham, Kara Combs, Lance M. Kaplan, Federico Cerutti 5/8/2026

Preliminary Insights in Chronos Frequency Data Understanding and Reconstruction

Analysis of Chronos foundation model's ability to process and represent frequency domain time-series data.

Ax Xin Peng, Ang Gao 5/8/2026

Flow Matching with Arbitrary Auxiliary Paths

Flow matching generative framework supporting arbitrary auxiliary distributions for flexible trajectories.

Ax Alexander Conzelmann, Albert Catalan-Tatjer, Shiwei Liu 5/8/2026

Layer Collapse in Diffusion Language Models

Analysis of layer collapse phenomenon in diffusion language models affecting activation dynamics.

Ax Hao Yu 5/8/2026

A Unified Pair-GRPO Family: From Implicit to Explicit Preference Constraints for Stable and General RL Alignment

Pair-GRPO framework for stable LLM alignment via reinforcement learning from human preferences.

Ax Ana Leticia Garcez Vicente, Gijs van Seeventer, Saber Salehkaleybar 5/8/2026

Data-Driven Covariate Selection for Nonparametric and Cycle-Agnostic Causal Effect Estimation

Data-driven local covariate selection method for causal effect estimation with latent confounding.

Ax Nan Jia, Haojin Yang, Xing Ma, Jiesong Lian, Shuailiang Zhang, Weipeng Zhang, Ke Zeng, Xunliang Cai, Zequn Sun 5/8/2026

Asymmetric On-Policy Distillation: Bridging Exploitation and Imitation at the Token Level

On-policy distillation method for LLM token-level training addressing variance and exploration issues.

Ax Kartik Tandon, Julian Gould, Tanishq Bhatia, Francesca Dominici, Alejandro Ribeiro, Claudio Battiloro 5/8/2026

Consistent Geometric Deep Learning via Hilbert Bundles and Cellular Sheaves

Research on unified convolutional learning framework for infinite-dimensional signals on manifolds.

Ax Liu Hanzuo, Chaofan Lin, Weixuan Sun, Yulong Wang, Key, Rayying, Mingyu Gao 5/8/2026

SparseForge: Efficient Semi-Structured LLM Sparsification via Annealing of Hessian-Guided Soft-Mask

Post-training semi-structured sparsification framework for LLMs using Hessian-guided soft masking and annealing for efficiency.

Ax Qingjun Zhang 5/8/2026

E = T*H/(O+B): A Dimensionless Control Parameter for Mixture-of-Experts Ecology

Dimensionless control parameter predicting mixture-of-experts model stability and expert ecology collapse across vision and language tasks.

Ax Daniel Uyterlinde, Swasti Shreya Mishra, Pascal Mettes 5/8/2026

Hyperbolic Concept Bottleneck Models

Interpretable concept bottleneck models using hyperbolic geometry to capture semantic hierarchies between concepts.

Ax Junye Du, Zhenghao Li, Yushi Feng, Long Feng 5/8/2026

FedFrozen: Two-Stage Federated Optimization via Attention Kernel Freezing

Federated learning optimization for Transformer models with attention kernel freezing to reduce client drift.

Ax Wenhan Zheng, Yuyi Mao, Ivan Wang-Hei Ho 5/8/2026

Scene-Adaptive Continual Learning for CSI-based Human Activity Recognition with Mixture of Experts

Continual learning approach for CSI-based activity recognition using mixture of experts to handle domain shifts.

Ax Maresa Schr\"oder, Pascal Janetzky, Michael Klar, Stefan Feuerriegel 5/8/2026

ORTHOBO: Orthogonal Bayesian Hyperparameter Optimization

Bayesian hyperparameter optimization addressing acquisition estimation noise and unstable candidate ranking decisions.

Ax Agnibh Dasgupta, Abdullah Tanvir, Xin Zhong 5/8/2026

Invariant Features in Language Models: Geometric Characterization and Model Attribution

Geometric framework characterizing invariant semantic features in language model latent space under paraphrasing.

Ax Weien Li, Rui Song, Zeyu Li, Haochen Liu, Gonghao Zhang, Difan Jiao, Zhenwei Tang, Bowei He, Haolun Wu, Xue Liu, Ye Yuan 5/8/2026

MINER: Mining Multimodal Internal Representation for Efficient Retrieval

Multimodal retrieval method mining internal representations for efficient visual document retrieval with reduced index footprint.

Ax Richard von Moos, Mathieu Alain, Bastian Rieck 5/8/2026

Invariant-Based Diagnostics for Graph Benchmarks

Graph invariant diagnostic framework for benchmarking graph foundation models to separate structure and feature contributions.

Ax Katharina Limbeck, Nadja H\"ausermann, Martin Carrasco, Guy Wolf, Bastian Rieck 5/8/2026

Diversity Curves for Graph Representation Learning

Graph representation learning method using diversity curves to compare graphs of different sizes with structural awareness.

Ax Johannes S. Schmidt, Martin Carrasco, Ernst R\"oell, Guy Wolf, Nello Blaser, Bastian Rieck 5/8/2026

No Triangulation Without Representation: Generalization in Topological Deep Learning

Benchmark evaluation framework for topological deep learning models on manifold data with triangulation operations.

Ax Haoyu Zheng, Fangcheng Fu, Jia Wu, Binhang Yuan, Yongqiang Zhang, Hao Wang, Yuanyuan Zhu, Xiao Yan, Jiawei Jiang 5/8/2026

Efficient Serving for Dynamic Agent Workflows with Prediction-based KV-Cache Management

arXiv paper on efficient LLM serving for dynamic agent workflows using prediction-based KV-Cache management.

Ax Xiang Li, Nan Jiang 5/8/2026

Q-MMR: Off-Policy Evaluation via Recursive Reweighting and Moment Matching

arXiv paper presenting Q-MMR framework for off-policy evaluation in MDPs via recursive reweighting and moment matching.

Ax Zuyuan Zhang, Fei Xu Yu, Tian Lan 5/8/2026

Operator-Guided Invariance Learning for Continuous Reinforcement Learning

arXiv paper on operator-guided invariance learning for continuous reinforcement learning under distribution shift.

Ax Chuanyang Zheng, Jiankai Sun, Yihang Gao, Yuehao Wang, Liangchen Tan, Mac Schwager, Anderson Schneider, Yuriy Nevmyvaka, Xiaodong Liu 5/8/2026

Cubit: Token Mixer with Kernel Ridge Regression

arXiv paper interpreting transformer attention as Nadaraya-Watson regression, proposing Cubit token mixer alternative.

Ax Murat Bilgehan Ertan, Xiaochen Zhu, Phuong Ha Nguyen, Marten van Dijk, Srinivas Devadas 5/8/2026

PACZero: PAC-Private Fine-Tuning of Language Models via Sign Quantization

arXiv paper on PACZero, a PAC-private zeroth-order method for fine-tuning LLMs via sign quantization.

Ax Amir Rezaei Balef, Mykhailo Koshil, Katharina Eggensperger 5/8/2026

Is One Layer Enough? Understanding Inference Dynamics in Tabular Foundation Models

arXiv paper analyzing layerwise inference dynamics in transformer-based tabular foundation models.