While recent continual learning methods largely alleviate the catastrophic problem on toy-sized datasets, some issues remain to be tackled to apply them to real-world problem domains. First, a continual learning model should effectively handle catastrophic forgetting and be efficient to train even with a large number of tasks. Secondly, it needs to tackle the problem of order-sensitivity, where the performance of the tasks largely varies based on the order of the task arrival sequence, as it may cause serious problems where fairness plays a critical role (e.g. medical diagnosis). To tackle these practical challenges, we propose a novel continual learning method that is scalable as well as order-robust, which instead of learning a completely shared set of weights, represents the parameters for each task as a sum of task-shared and sparse task-adaptive parameters. With our Additive Parameter Decomposition (APD), the task-adaptive parameters for earlier tasks remain mostly unaffected, where we update them only to reflect the changes made to the task-shared parameters. This decomposition of parameters effectively prevents catastrophic forgetting and order-sensitivity, while being computation- and memory-efficient. Further, we can achieve even better scalability with APD using hierarchical knowledge consolidation, which clusters the task-adaptive parameters to obtain hierarchically shared parameters. We validate our network with APD, APD-Net, on multiple benchmark datasets against state-of-the-art continual learning methods, which it largely outperforms in accuracy, scalability, and order-robustness.
Modular Delta Merging with Orthogonal Constraints: A Scalable Framework for Continual and Reversible Model Composition
Rensselaer Polytechnic Institute(伦斯勒理工学院);National University of Sciences and Technology (NUST)(国立科技大学);Virginia Tech(弗吉尼亚理工大学);University of New Mexico(新墨西哥大学)
Learn to Memorize: Scalable Continual Learning in Semiparametric Models with Mixture-of-Neighbors Induction Memory
State Key Laboratory for Multimedia Information Processing, School of Computer Science, Peking University(多媒体信息处理国家重点实验室,计算机科学学院,北京大学);Microsoft(微软公司);Microsoft Research Asia(微软亚洲研究院)
Building a Subspace of Policies for Scalable Continual Learning
Meta AI Research(Meta人工智能研究院);CNRS-ISIR, Sorbonne University(法国国家科学研究中心智能系统与机器人研究所,索邦大学);McGill University(麦吉尔大学);Mila(米拉人工智能研究所);Bosch Research(博世研究院);Ubisoft France(育碧法国)
Compacting, Picking and Growing for Unforgetting Continual Learning
Institute of Information Science, Academia Sinica(中央研究院信息科学研究所);MOST Joint Research Center for AI Technology and All Vista Healthcare(科技部人工智能技术与全视角医疗联合研究中心)
Improving Continual Learning Performance and Efficiency with Auxiliary Classifiers
Warsaw University of Technology(华沙技术大学);Computer Vision Center, Barcelona(巴塞罗那计算机视觉中心);IDEAS Research Institute(IDEAS研究机构);Institute of Artificial Intelligence and Robotics(人工智能与机器人研究所);Xi’an Jiaotong University(西安交通大学);Nankai University(南开大学)
Exploring the Impact of Parameter Update Magnitude on Forgetting and Generalization of Continual Learning
Institute of Intelligent Information Processing, Shanxi University(山西大学智能信息处理研究所);Alliance Manchester Business School, The University of Manchester(曼彻斯特大学阿利安斯商学院)
Caption, Create, Continue: Continual Learning with Pre-trained Generative Vision-Language Models
International Institute of Information Technology Bangalore (IIITB), India(国际信息技术研究所(班加罗尔));Institute for Infocomm Research, Agency for Science, Technology and Research (A*STAR), Singapore(信息与通信研究所(A*STAR))
The Future of Continual Learning in the Era of Foundation Models: Three Key Directions
Department of Computer Science, Università di Pisa(帕尔马大学计算机科学系);Institute of Information Science and Technologies, National Research Council(信息科学与技术研究所,国家研究理事会)
M2RU: Memristive Minion Recurrent Unit for On-Chip Continual Learning at the Edge
NUAI Lab, Department of Electrical and Computer Engineering, University of Texas at San Antonio and Department of Electrical Engineering, University of Baghdad(NUAI实验室,电子与计算机工程系,德克萨斯大学圣安东尼奥分校和电子工程系,巴格达大学)
GRID: Scaling Task-Agnostic Inference in Continual Prompt Tuning
State University of New York at Buffalo(纽约州立大学布法罗分校);Department of Computer Science and Engineering(计算机科学与工程系);Institute for Artificial Intelligence and Data Science(人工智能与数据科学研究院)
SD-LoRA: Scalable Decoupled Low-Rank Adaptation for Class Incremental Learning
City University of Hong Kong(香港城市大学);Harvard University(哈佛大学);Xi’an Jiaotong University(西安交通大学);Tencent AI Lab(腾讯AI实验室);Zhejiang University(浙江大学);Pengcheng Laboratory(鹏城实验室)