Publications

Position: Modular Memory is the Key to Continual Learning Agents
Vaggelis Dorovatas
Malte Schwerin
Andrew Bagdanov
Lucas Caccia
Antonio Carta
Laurent Charlin
CITEC Barbara Hammer
Tyler Hayes
Timm Hess
Christopher Kanan
Dhireesha Kudithipudi
Xialei Liu
Vincenzo Lomonaco
Jorge Mendez-Mendez
Darshan Patil
Ameya Pandurang Prabhu
Elisa Ricci
Tinne Tuytelaars
Gido van de Ven
Liyuan Wang … (see 4 more)
Joost van de Weijer
Jonghyun Choi
Martin Mundt
Rahaf Aljundi
Foundation models have transformed machine learning through large-scale pretraining, massive parameterization, and increased test-time compu… (see more)te. Despite surpassing human performance in several domains, these models remain fundamentally limited in continuous operation, experience accumulation, and personalization, capabilities that are central to adaptive intelligence. While continual learning research has long targeted these goals, its historical focus on in-weight learning, i.e., updating a single model’s parameters to absorb new knowledge, has rendered catastrophic forgetting a persistent challenge. **Our position is that combining the strengths of In-Weight Learning (IWL) and the newly emerged capabilities of In-Context Learning (ICL) through the design of modular memory is the missing piece for continual adaptation at scale.** We outline a conceptual framework for modular memory-centric architectures that leverage ICL for rapid adaptation and knowledge accumulation, and IWL for stable updates to model capabilities, thereby mitigating catastrophic forgetting and charting a practical roadmap toward continually learning agents.
2025-12-31
International Conference on Machine Learning (Accept (spotlight))
openreview.net
openreview.net
Position: Preparing for AI Systems That Deceive Developers
Isabella Duan
Xudong Pan
Yawen Duan
Adam Gleave
Ranjie Duan
Jianfeng Cao
Wenqi Chen
Yinpeng Dong
Jiarun Dai
Jie Fu
Xudong Guo
Tianxing He
Geng Hong
Naying HU
Xiaojian Li
Dongrui Liu
Chaochao Lu
Sören Mindermann
Peng Xu
Yang Zhang … (see 4 more)
Chen Zheng
Brian Tse
Min Yang
Xia Hu
AI systems may exhibit deceptive behaviors that mislead developers about their capabilities, propensities, or actions. Such deception can ta… (see more)ke distinct forms across the development lifecycle: training subversion, evaluation gaming, and control evasion. We argue that the AI community should prioritize AI deception targeting developers as a distinct risk category because it compromises developers' ability to identify and mitigate all other risks. We propose three recommendations for developers: preserving monitorability during training, ensuring safety evaluation integrity against evaluation-aware systems, and establishing non-evadable control prior to deployment. We identify open problems for the research community, whose resolution is critical for the safe development of frontier AI.
2025-12-31
International Conference on Machine Learning (Accept (regular))
openreview.net
openreview.net
Position: Time to Close The Validation Gap in LLM Social Simulations
Maximilian Puelma Touzel
Sneheel Sarangi
Aurélien Bück-Kaeffer
Zachary Yang
Jean-François Godbout
Reihaneh Rabbany
LLM-based social simulations—in which many language model agents interact over multiple turns—are rapidly proliferating across policy an… (see more)alysis, epidemiology, and computational social science. Yet the field lacks consensus on how to validate these simulations, with evaluation methods that are sparse, inconsistent, and rarely shared across disciplinary silos. We argue this creates a serious risk: premature deployment of unvalidated simulators in high-stakes domains. Our position is that the field must pivot from expansion to consolidation, prioritizing methodological standardization—shared benchmarks, open data, and reproducible evaluation protocols grounded in social science and complex systems research. We outline a concrete research program organized around specific learning problems/benchmarks, providing a path toward answering the fundamental question: when are LLM social simulations useful modelling objects?
2025-12-31
International Conference on Machine Learning (Accept (regular))
openreview.net
openreview.net
Practical Solutions to Volt-var Optimization under Uncertainty via Blackbox Optimization
Feng Li
Ilhan Kocar
Antoine Lesage-Landry
In this work, we propose an optimal reactive power dispatch (ORPD) stochastic program for volt-var optimization (VVO) of power distribution … (see more)networks. The formulation considers not only control settings of conventional VVO devices, e.g., voltage regulators, capacitor banks, and on-load tap changers, but also optimal settings for volt-var droop curves of distributed energy resources (DERs), compliant with the IEEE 1547-2018 standard. Instead of including the power flow equations in the optimization problem which makes it nonlinear and nonconvex, a power flow solver is utilized and the problem is solved by blackbox optimization (BBO). The feasibility of the derived solution is improved by using unbalanced power flow simulations. The solution is effective under various demand and DER generation scenarios such that device settings are not frequently changed, making it practical for in-field implementations. Through numerical simulations on IEEE test feeders, we illustrate the performance of the solutions of our proposed approach on both in-sample and out-of-sample scenarios. We show that our approach outperforms a benchmark reinforcement learning method, and is also scalable to large-scale distribution networks.
2025-12-31
IEEE Transactions on Power Delivery (published)
doi.org
Prompt learning with bounding box constraints for medical image segmentation.
Melanie Gaillochet
Mehrdad Noori
Sahar Dastani
Christian Desrosiers
Hervé Lombaert
Pixel-wise annotations are notoriously labourious and costly to obtain in the medical domain. To mitigate this burden, weakly supervised app… (see more)roaches based on bounding box annotations-much easier to acquire-offer a practical alternative. Vision foundation models have recently shown noteworthy segmentation performance when provided with prompts such as points or bounding boxes. Prompt learning exploits these models by adapting them to downstream tasks and automating segmentation, thereby reducing user intervention. However, existing prompt learning approaches depend on fully annotated segmentation masks. This paper proposes a novel framework that combines the representational power of foundation models with the annotation efficiency of weakly supervised segmentation. More specifically, our approach automates prompt generation for foundation models using only bounding box annotations. Our proposed optimization scheme integrates multiple constraints derived from box annotations with pseudo-labels generated by the prompted foundation model. Extensive experiments across multi-modal datasets reveal that our weakly supervised method achieves an average Dice score of 84.90% in a limited data setting, outperforming existing fully-supervised and weakly-supervised approaches. The code will be available upon acceptance
2025-12-31
IEEE Trans. Biomed. Eng. (published)
doi.org
arxiv.org
Property-Driven Protein Inverse Folding with Multi-Objective Preference Alignment
Junqi Liu
Xiaoyang Hou
Chence Shi
Xin Liu
Zhi Yang
Jian Tang
Protein sequence design must balance designability, defined as the ability to recover a target backbone, with multiple, often competing, dev… (see more)elopability properties such as solubility, thermostability, and expression. Existing approaches address these properties through post hoc mutation, inference-time biasing, or retraining on property-specific subsets, yet they are target dependent and demand substantial domain expertise or careful hyperparameter tuning. In this paper, we introduce ProtAlign, a multi-objective preference alignment framework that fine-tunes pretrained inverse folding models to satisfy diverse developability objectives while preserving structural fidelity. ProtAlign employs a semi-online Direct Preference Optimization strategy with a flexible preference margin to mitigate conflicts among competing objectives and constructs preference pairs using in silico property predictors. Applied to the widely used ProteinMPNN backbone, the resulting model MoMPNN enhances developability without compromising designability across tasks including sequence design for CATH 4.3 crystal structures, de novo generated backbones, and real-world binder design scenarios, making it an appealing framework for practical protein sequence design.
2025-12-31
International Conference on Learning Representations (Accept (Poster))
doi.org
openreview.net
Quantifying LLM Attention-Head Stability: Implications for Circuit Universality.
Karan Bali
Jack Stanley
Praneet Suresh
Danilo Bzdok
In mechanistic interpretability, recent work scrutinizes transformer"circuits"- sparse, mono or multi layer sub computations, that may refle… (see more)ct human understandable functions. Yet, these network circuits are rarely acid-tested for their stability across different instances of the same deep learning architecture. Without this, it remains unclear whether reported circuits emerge universally across labs or turn out to be idiosyncratic to a particular estimation instance, potentially limiting confidence in safety-critical settings. Here, we systematically study stability across-refits in increasingly complex transformer language models of various sizes. We quantify, layer by layer, how similarly attention heads learn representations across independently initialized training runs. Our rigorous experiments show that (1) middle-layer heads are the least stable yet the most representationally distinct; (2) deeper models exhibit stronger mid-depth divergence; (3) unstable heads in deeper layers become more functionally important than their peers from the same layer; (4) applying weight decay optimization substantially improves attention-head stability across random model initializations; and (5) the residual stream is comparatively stable. Our findings establish the cross-instance robustness of circuits as an essential yet underappreciated prerequisite for scalable oversight, drawing contours around possible white-box monitorability of AI systems.
2025-12-31
arXiv (preprint)
doi.org
arxiv.org
RAEE: A Robust Retrieval-Augmented Early Exit Framework for Efficient Inference
Lianming Huang
Shangyu Wu
Yufei Cui
Ying Xiong
Haibo Hu
Xue Liu
Tei-Wei Kuo
Nan Guan
Chun Jason Xue
Deploying large language model inference remains challenging due to their high computational overhead. Early exit optimizes model inference … (see more)by adaptively reducing the number of inference layers. Current methods typically train internal classifiers or use heuristic methods to determine the exit layer. However, those methods either introduce significant training overheads or lead to performance degradation. To address these limitations, this paper proposes RAEE, a robust Retrieval-Augmented Early Exit framework that not only enables early exit but also enhances model performance through corrective exit information at intermediate layers. This paper first demonstrates that the early exit problem can be effectively modeled as a distribution prediction problem, in which the distribution can be further approximated through the exit information of similar data. Subsequently, this paper introduces the process of collecting exit information of correct predictions and the steps to construct the retrieval database. Finally, leveraging the pre-constructed retrieval database, RAEE utilizes the exit information from retrieved similar data to guide the backbone model's exit. Experimental results demonstrate that RAEE can not only accelerate inference while achieving robust zero-shot performance across eight downstream tasks.
2025-12-31
International Conference on Learning Representations (Accept (Poster))
doi.org
openreview.net
Recall, Robustness, and Lexicographic Evaluation
Fernando Diaz
Michael D. Ekstrand
Bhaskar Mitra
2025-12-31
Trans. Recomm. Syst. (published)
doi.org
arxiv.org
RECODE: A Benchmark for Research Code DEvelopment with Interactive Human Feedback
Chunyu Miao
Henry Peng Zou
Yangning Li
Yankai Chen
Yibo Wang
Fangxin Wang
Yifan Li
Wooseong Yang
Bowei He
Xinni Zhang
Dianzhi Yu
Hanchen Yang
Hoang H Nguyen
Yue Zhou
Jie Yang
Jizhou Guo
Wenzhe Fan
Chin-Yuan Yeh
Panpan Meng
Liancheng Fang … (see 11 more)
Jinhu Qi
Wei-Chieh Huang
Zhengyao Gu
Yuwei Han
Langzhou He
Yuyao Yang
Yinghui Li
Hai-Tao Zheng
Xue Liu
Irwin King
Philip S. Yu
Large language models (LLMs) show the promise in supporting scientific research implementation, yet their ability to generate correct and ex… (see more)ecutable code remains limited. Existing works largely adopt one-shot settings, ignoring the iterative and feedback-driven nature of realistic workflows of scientific research development. To address this gap, we present RECODE, a benchmark of 102 tasks from research papers and repositories that evaluates LLMs through multi-turn interactions with human feedback. It includes structured instructions, unit tests, and a five-level feedback hierarchy to reflect realistic researcher–agent collaboration. We further present ReCodeAgent, a framework that integrates feedback into iterative code generation. Experimentswith leading LLMs, including GPT-5, Claude-Sonnet-4, DeepSeek-V3.1, and Gemini 2.5, show substantial performance gains with richer feedback, while also highlighting ongoing challenges in the generation of complex research code. RECODE establishes a foundation for developing adaptive, feedback-driven LLM agents in scientific research implementation.
2025-12-31
International Conference on Learning Representations (Accept (Poster))
openreview.net
openreview.net
Recurrent Structural Policy Gradient for Partially Observable Mean Field Games
Clarisse Wibault
Johannes Forkel
Sebastian Towers
Tiphaine Wibault
Juan Duque
George Whittle
Andreas Schaab
Yucheng Yang
Chiyuan Wang
Maike Osborne
Benjamin Moll
Jakob Foerster
Mean Field Games (MFGs) provide a principled framework for modeling interactions in large populations models: at scale, population dynamics … (see more)become deterministic, with uncertainty entering only through aggregate shocks, or *common noise*. However, algorithmic progress has been limited since model-free methods are too high variance and exact methods scale poorly. Recent Hybrid Structural Methods (HSMs) use Monte Carlo rollouts for the common noise in combination with exact estimation of the expected return, conditioned on those samples. However, HSMs have not been scaled to Partially Observable settings. We propose *Recurrent Structural Policy Gradient* (RSPG), the first history-aware HSM. We also introduce MFAX, our JAX-based framework for MFGs. By leveraging known transition dynamics, RSPG achieves state-of-the-art performance as well as an order-of-magnitude faster convergence and solves, for the first time, a macroeconomics MFG with heterogeneous agents, common noise and history-aware policies. MFAX is publicly available at: .
2025-12-31
International Conference on Machine Learning (Accept (spotlight))
doi.org
openreview.net
Reinforced Sequential Monte Carlo for Amortised Sampling
Sanghyeok Choi
Sarthak Mittal
Víctor Elvira
Jinkyoo Park
Esmeralda S. Whitammer
This paper proposes a synergy of amortised and particle-based methods for sampling from distributions defined by unnormalised density functi… (see more)ons. We state a connection between sequential Monte Carlo (SMC) and neural sequential samplers trained by maximum-entropy reinforcement learning (MaxEnt RL), wherein learnt sampling policies and value functions define proposal kernels and twist functions. Exploiting this connection, we introduce an off-policy RL training procedure for the sampler that uses samples from SMC -- using the learnt sampler as a proposal -- as a behaviour policy that better explores the target distribution. We describe techniques for stable joint training of proposals and twist functions and an adaptive weight tempering scheme to reduce training signal variance. Furthermore, building upon past attempts to use experience replay to guide the training of neural samplers, we derive a way to combine historical samples with annealed importance sampling weights within a replay buffer. On synthetic multi-modal targets (in both continuous and discrete spaces) and the Boltzmann distribution of alanine dipeptide conformations, we demonstrate improvements in approximating the true distribution as well as training stability compared to both amortised and Monte Carlo methods.
2025-12-31
International Conference on Machine Learning (Accept (spotlight))
openreview.net
openreview.net