LAMP

Category: Publications

  • Online Continual Learning with Dynamic Label Hierarchies

    —

    published in

    Xinrui Wang, Shao-Yuan Li, Bartłomiej Twardowski, Alexandra Gomez-Villa, Songcan Chen Read Full Paper → Online Continual Learning (OCL) aims to learn from endless non\text{-}stationary data streams, yet most existing methods assume a flat label space and overlook the hierarchical organization of real\text{-}world concepts that evolves both horizontally (sibling classes) and vertically (coarse or fine categories). To better reflect this…

  • Less Precise Can Be More Reliable: A Systematic Evaluation of Quantization’s Impact on VLMs Beyond Accuracy

    —

    published in

    Aymen Bouguerra, Daniel Montoya, Alexandra Gomez-Villa, Chokri Mraidha, Fabio Arnez Read Full Paper → Vision-Language Models (VLMs) such as CLIP have revolutionized zero-shot classification and safety-critical tasks, including Out-of-Distribution (OOD) detection. However, their high computational cost hinders efficient real-world deployment. While quantization is a standard solution for efficiency, its broader impact on reliability metrics beyond simple Top-1 accuracy remains…

  • Modular Memory is the Key to Continual Learning Agents

    —

    published in

    Vaggelis Dorovatas, Malte Schwerin, Andrew D. Bagdanov, Lucas Caccia, Antonio Carta, Laurent Charlin, Barbara Hammer, Tyler L. Hayes, Timm Hess, Christopher Kanan, Dhireesha Kudithipudi, Xialei Liu, Vincenzo Lomonaco, Jorge Mendez-Mendez, Darshan Patil, Ameya Prabhu, Elisa Ricci, Tinne Tuytelaars, Gido M. van de Ven, Liyuan Wang, Joost van de Weijer, Jonghyun Choi, Martin Mundt, Rahaf Aljundi Read Full Paper → Foundation models have transformed machine learning through large-scale pretraining and increased test-time compute. Despite surpassing human performance in several…

  • Towards Dynamic Modality Alignment in Multimodal Continual Learning

    —

    published in

    Jiayao Tan ⋅ Fan Lyu ⋅ Tianle Liu ⋅ Fuyuan Hu ⋅ Wei Feng Read Full Paper → Multimodal Continual Learning (MMCL) aims to enable models to continuously accumulate knowledge across multiple tasks and modalities without forgetting prior information. MMCL presents more challenges than single-modal continual learning, as it requires effective cooperation and complementarity between…

  • Subspace Alignment for CLIP-based Continual Learning via Canonical Correlation Analysis

    —

    published in

    Huan Zhang ⋅ Shuyu Dong ⋅ Yujin Zheng ⋅ Dingwen Wang ⋅ Shenghua Fan ⋅ Fan Lyu Read Full Paper → Recent advances in CLIP-based continual learning have shown the potential of leveraging pre-trained vision-language models for sequential tasks. However, existing methods overlook a key problem we call Asymmetric Drift. In unimodal CLIP-based continual learning,…

  • GenColorBench: A Color Evaluation Benchmark for Text-to-Image Generation

    —

    published in

    Muhammad Atif Butt, Alexandra Gomez-Villa, Tao Wu, Javier Vazquez-Corral, Joost Van De Weijer, Kai Wang Read Full Paper → Recent years have seen impressive advances in text-to-image generation, with image generative or unified models, generating high-quality images from text. Yet these models still struggle with fine-grained color control, often failing to accurately match colors specified in text prompts. While existing…

  • IsoCLIP: Decomposing CLIP Projectors for Efficient Intra-modal Alignment

    —

    published in

    Simone Magistri, Dipam Goswami, Marco Mistretta, Bartłomiej Twardowski, Joost van de Weijer, Andrew D. Bagdanov Read Full Paper → Vision-Language Models like CLIP are extensively used for inter-modal tasks which involve both visual and text modalities. However, when the individual modality encoders are applied to inherently intra-modal tasks like image-to-image retrieval, their performance suffers from the intra-modal misalignment. In this…

  • Revisiting Weight Regularization for Low-Rank Continual Learning

    —

    published in

    Yaoyue Zheng, Yin Zhang, Joost van de Weijer, Gido M van de Ven, Shaoyi Du, Xuetao Zhang, Zhiqiang Tian Read Full Paper → Continual Learning (CL) with large-scale pre-trained models (PTMs) has recently gained wide attention, shifting the focus from training from scratch to continually adapting PTMs. This has given rise to a promising paradigm: parameter-efficient continual learning (PECL), where task…

  • From Cradle to Cane: A Two-Pass Framework for High-Fidelity Lifespan Face Aging

    —

    published in

    Tao Liu, Dafeng Zhang, Gengchen Li, Shizhuo Liu, Yongqi Song, Senmao Li, Shiqi Yang, Boqian Li, Kai Wang, Yaxing Wang Read Full Paper → Face aging has become a crucial task in computer vision, with applications ranging from entertainment to healthcare. However, existing methods struggle with achieving a realistic and seamless transformation across the entire lifespan, especially when handling large age gaps or extreme…

  • Accurate and Efficient Low-Rank Model Merging in Core Space

    —

    published in

    Aniello Panariello, Daniel Marczak, Simone Magistri, Angelo Porrello, Bartłomiej Twardowski, Andrew D. Bagdanov, Simone Calderara, Joost van de Weijer Read Full Paper → In this paper, we address the challenges associated with merging low-rank adaptations of large neural networks. With the rise of parameter-efficient adaptation techniques, such as Low-Rank Adaptation (LoRA), model fine-tuning has become more accessible. While fine-tuning models with LoRA…