Continual Learning (CL) epitomizes an advanced training paradigm wherein prior data samples remain inaccessible during the acquisition of new tasks. Numerous investigations have delved into leveraging a pre-trained Vision Transformer (ViT) to enhance model efficacy in continual learning. Nonetheless, these approaches typically utilize a singular, static backbone, which inadequately adapts to novel tasks, particularly when engaging with diverse data domains, due to a substantial number of inactive parameters. This paper addresses this limitation by introducing an innovative Self-Controlled Dynamic Expansion Model (SCDEM), which orchestrates multiple distinct trainable pre-trained ViT backbones to furnish diverse and semantically enriched representations. Specifically, by employing the multi-backbone architecture as a shared module, the proposed SCDEM dynamically generates a new expert with minimal parameters to accommodate a new task. A novel Collaborative Optimization Mechanism (COM) is introduced to synergistically optimize multiple backbones by harnessing prediction signals from historical experts, thereby facilitating new task learning without erasing previously acquired knowledge. Additionally, a novel Feature Distribution Consistency (FDC) approach is proposed to align semantic similarity between previously and currently learned representations through an optimal transport distance-based mechanism, effectively mitigating negative knowledge transfer effects. Furthermore, to alleviate over-regularization challenges, this paper presents a novel Dynamic Layer-Wise Feature Attention Mechanism (DLWFAM) to autonomously determine the penalization intensity on each trainable representation layer. An extensive series of experiments have been conducted to evaluate the proposed methodology's efficacy, with empirical results corroborating that the approach attains state-of-the-art performance.
Objectives: Machine learning (ML) has the potential to facilitate "continual learning" in medicine, in which an ML system continues to evolve in response to exposure to new data over time, even after being deployed in a clinical setting. In this paper, we provide a tutorial on the range of ethical issues raised by the use of such "adaptive" ML systems in medicine that have, thus far, been neglected in the literature. Target audience: The target audiences for this tutorial are the developers of machine learning AI systems, healthcare regulators, the broader medical informatics community, and practicing clinicians. Scope: Discussions of adaptive ML systems to date have overlooked the distinction between two sorts of variance that such systems may exhibit -- diachronic evolution (change over time) and synchronic variation (difference between cotemporaneous instantiations of the algorithm at different sites) -- and under-estimated the significance of the latter. We highlight the challenges that diachronic evolution and synchronic variation present for the quality of patient care, informed consent, and equity, and discuss the complex ethical trade-offs involved in the design of such systems.
The integration of advanced sensor technologies with deep learning algorithms has revolutionized fault diagnosis in railway systems, particularly at the wheel-track interface. Although numerous models have been proposed to detect irregularities such as wheel out-of-roundness, they often fall short in real-world applications due to the dynamic and nonstationary nature of railway operations. This paper introduces BOLT-RM (Boosting-inspired Online Learning with Transfer for Railway Maintenance), a model designed to address these challenges using continual learning for predictive maintenance. By allowing the model to continuously learn and adapt as new data become available, BOLT-RM overcomes the issue of catastrophic forgetting that often plagues traditional models. It retains past knowledge while improving predictive accuracy with each new learning episode, using a boosting-like knowledge sharing mechanism to adapt to evolving operational conditions such as changes in speed, load, and track irregularities. The methodology is validated through comprehensive multi-domain simulations of train-track dynamic interactions, which capture realistic railway operating conditions. The proposed BOLT-RM model demonstrates significant improvements in identifying wheel anomalies, establishing a reliable sequence for maintenance interventions.
As generative AI image technologies become more widespread and advanced, there is a growing need for strong attribution models. These models are crucial for verifying the authenticity of images and identifying the architecture of their originating generative models-key to maintaining media integrity. However, attribution models struggle to generalize to unseen models, and traditional fine-tuning methods for updating these models have shown to be impractical in real-world settings. To address these challenges, we propose LoRA eXpandable Networks (LoRAX), a parameter-efficient class incremental algorithm that adapts to novel generative image models without the need for full retraining. Our approach trains an extremely parameter-efficient feature extractor per continual learning task via Low Rank Adaptation. Each task-specific feature extractor learns distinct features while only requiring a small fraction of the parameters present in the underlying feature extractor's backbone model. Our extensive experimentation shows LoRAX outperforms or remains competitive with state-of-the-art class incremental learning algorithms on the Continual Deepfake Detection benchmark across all training scenarios and memory settings, while requiring less than 3% of the number of trainable parameters per feature extractor compared to the full-rank implementation. LoRAX code is available at: https://github.com/mit-ll/lorax_cil.
Continual generalized category discovery has been introduced and studied in the literature as a method that aims to continuously discover and learn novel categories in incoming data batches while avoiding catastrophic forgetting of previously learned categories. A key component in addressing this challenge is the model's ability to separate novel samples, where Extreme Value Theory (EVT) has been effectively employed. In this work, we propose a novel method that integrates EVT with proxy anchors to define boundaries around proxies using a probability of inclusion function, enabling the rejection of unknown samples. Additionally, we introduce a novel EVT-based loss function to enhance the learned representation, achieving superior performance compared to other deep-metric learning methods in similar settings. Using the derived probability functions, novel samples are effectively separated from previously known categories. However, category discovery within these novel samples can sometimes overestimate the number of new categories. To mitigate this issue, we propose a novel EVT-based approach to reduce the model size and discard redundant proxies. We also incorporate experience replay and knowledge distillation mechanisms during the continual learning stage to prevent catastrophic forgetting. Experimental results demonstrate that our proposed approach outperforms state-of-the-art methods in continual generalized category discovery scenarios.
As edge devices become more capable and pervasive in wireless networks, there is growing interest in leveraging their collective compute power for distributed learning. However, optimizing learning at the network edge entails unique challenges, particularly when moving beyond conventional settings and objectives. While Federated Learning (FL) has emerged as a key paradigm for distributed model training, critical challenges persist. First, existing approaches often overlook the trade-off between predictive accuracy and interpretability. Second, they struggle to integrate inherently explainable models such as decision trees because their non-differentiable structure makes them not amenable to backpropagation-based training algorithms. Lastly, they lack meaningful mechanisms for continual Machine Learning (ML) model adaptation through Continual Learning (CL) in resource-limited environments. In this paper, we pave the way for a set of novel optimization problems that emerge in distributed learning at the network edge with wirelessly interconnected edge devices, and we identify key challenges and future directions. Specifically, we discuss how Multi-objective optimization (MOO) can be used to address the trade-off between predictive accuracy and explainability when using complex predictive models. Next, we discuss the implications of integrating inherently explainable tree-based models into distributed learning settings. Finally, we investigate how CL strategies can be effectively combined with FL to support adaptive, lifelong learning when limited-size buffers are used to store past data for retraining. Our approach offers a cohesive set of tools for designing privacy-preserving, adaptive, and trustworthy ML solutions tailored to the demands of edge computing and intelligent services.
Algorithms and approaches for continual reinforcement learning have gained increasing attention. Much of this early progress rests on the foundations and standard practices of traditional reinforcement learning, without questioning if they are well-suited to the challenges of continual learning agents. We suggest that many core foundations of traditional RL are, in fact, antithetical to the goals of continual reinforcement learning. We enumerate four such foundations: the Markov decision process formalism, a focus on optimal policies, the expected sum of rewards as the primary evaluation metric, and episodic benchmark environments that embrace the other three foundations. Shedding such sacredly held and taught concepts is not easy. They are self-reinforcing in that each foundation depends upon and holds up the others, making it hard to rethink each in isolation. We propose an alternative set of all four foundations that are better suited to the continual learning setting. We hope to spur on others in rethinking the traditional foundations, proposing and critiquing alternatives, and developing new algorithms and approaches enabled by better-suited foundations.
Continual learning empowers models to learn from a continuous stream of data while preserving previously acquired knowledge, effectively addressing the challenge of catastrophic forgetting. In this study, we propose a new approach that integrates adapters within the self-attention mechanisms of Vision Transformers to enhance knowledge retention when sequentially adding datasets from different domains. Unlike previous methods that continue learning with only one dataset, our approach introduces domain-specific output heads and feature gating, allowing the model to maintain high accuracy on previously learned tasks while incorporating only the essential information from multiple domains. The proposed method is compared to prominent parameter-efficient fine-tuning methods in the current state of the art. The results provide evidence that our method effectively alleviates the limitations of previous works. Furthermore, we conduct a comparative analysis using three datasets, CIFAR-100, Flowers102, and DTD, each representing a distinct domain, to investigate the impact of task order on model performance. Our findings underscore the critical role of dataset sequencing in shaping learning outcomes, demonstrating that strategic ordering can significantly improve the model's ability to adapt to evolving data distributions over time while preserving the integrity of previously learned knowledge.
Neural Fields (NF) have gained prominence as a versatile framework for complex data representation. This work unveils a new problem setting termed \emph{Meta-Continual Learning of Neural Fields} (MCL-NF) and introduces a novel strategy that employs a modular architecture combined with optimization-based meta-learning. Focused on overcoming the limitations of existing methods for continual learning of neural fields, such as catastrophic forgetting and slow convergence, our strategy achieves high-quality reconstruction with significantly improved learning speed. We further introduce Fisher Information Maximization loss for neural radiance fields (FIM-NeRF), which maximizes information gains at the sample level to enhance learning generalization, with proved convergence guarantee and generalization bound. We perform extensive evaluations across image, audio, video reconstruction, and view synthesis tasks on six diverse datasets, demonstrating our method's superiority in reconstruction quality and speed over existing MCL and CL-NF approaches. Notably, our approach attains rapid adaptation of neural fields for city-scale NeRF rendering with reduced parameter requirement.
Continual learning techniques employ simple replay sample selection processes and use them during subsequent tasks. Typically, they rely on labeled data. In this paper, we depart from this by automatically selecting prototypes stored without labels, preserving cluster structures in the latent space across tasks. By eliminating label dependence in the replay buffer and introducing cluster preservation loss, it is demonstrated that the proposed method can maintain essential information from previously encountered tasks while ensuring adaptation to new tasks. "Push-away" and "pull-toward" mechanisms over previously learned prototypes are also introduced for class-incremental and domain-incremental scenarios. These mechanisms ensure the retention of previously learned information as well as adaptation to new classes or domain shifts. The proposed method is evaluated on several benchmarks, including SplitCIFAR100, SplitImageNet32, SplitTinyImageNet, and SplitCaltech256 for class-incremental, as well as R-MNIST and CORe50 for domain-incremental setting using pre-extracted DINOv2 features. Experimental results indicate that the label-free replay-based technique outperforms state-of-the-art continual learning methods and, in some cases, even surpasses offline learning. An unsupervised variant of the proposed technique for the class-incremental setting, avoiding labels use even on incoming data, also demonstrated competitive performance, outperforming particular supervised baselines in some cases. These findings underscore the effectiveness of the proposed framework in retaining prior information and facilitating continual adaptation.
Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models.
Low-Rank Adaptation (LoRA) has emerged as a popular parameter-efficient fine-tuning (PEFT) method for Large Language Models (LLMs), yet it still incurs notable overhead and suffers from parameter interference in multi-task scenarios. We propose LoRA with Reduced Interference (LoRI), a simple yet effective approach that freezes the projection matrices $A$ as random projections and sparsifies the matrices $B$ using task-specific masks. This design substantially reduces the number of trainable parameters while maintaining strong task performance. Moreover, LoRI minimizes cross-task interference in adapter merging by leveraging the orthogonality between adapter subspaces, and supports continual learning by using sparsity to mitigate catastrophic forgetting. Extensive experiments across natural language understanding, mathematical reasoning, code generation, and safety alignment tasks demonstrate that LoRI outperforms full fine-tuning and existing PEFT methods, while using up to 95% fewer trainable parameters than LoRA. In multi-task experiments, LoRI enables effective adapter merging and continual learning with reduced cross-task interference. Code is available at: https://github.com/juzhengz/LoRI
This paper introduces a self-learning agent that integrates LLaMA 3.2 with a Progressive Neural Network (PNN) for continual learning in conversational AI and code generation. The framework dynamically collects data, fine-tunes tasks with minimal samples, and leverages Meta-Learning for rapid adaptation. LoRA optimizes fine-tuning, while Elastic Weight Consolidation (EWC) enhances knowledge retention. Experimental results demonstrate improved adaptability and memory stability, positioning this approach as a scalable step toward Artificial General Intelligence (AGI).
The advent of large language models (LLMs) has catalyzed a transformative shift in artificial intelligence, paving the way for advanced intelligent agents capable of sophisticated reasoning, robust perception, and versatile action across diverse domains. As these agents increasingly drive AI research and practical applications, their design, evaluation, and continuous improvement present intricate, multifaceted challenges. This survey provides a comprehensive overview, framing intelligent agents within a modular, brain-inspired architecture that integrates principles from cognitive science, neuroscience, and computational research. We structure our exploration into four interconnected parts. First, we delve into the modular foundation of intelligent agents, systematically mapping their cognitive, perceptual, and operational modules onto analogous human brain functionalities, and elucidating core components such as memory, world modeling, reward processing, and emotion-like systems. Second, we discuss self-enhancement and adaptive evolution mechanisms, exploring how agents autonomously refine their capabilities, adapt to dynamic environments, and achieve continual learning through automated optimization paradigms, including emerging AutoML and LLM-driven optimization strategies. Third, we examine collaborative and evolutionary multi-agent systems, investigating the collective intelligence emerging from agent interactions, cooperation, and societal structures, highlighting parallels to human social dynamics. Finally, we address the critical imperative of building safe, secure, and beneficial AI systems, emphasizing intrinsic and extrinsic security threats, ethical alignment, robustness, and practical mitigation strategies necessary for trustworthy real-world deployment.
Meta-Continual Learning (Meta-CL) has emerged as a promising approach to minimize manual labeling efforts and system resource requirements by enabling Continual Learning (CL) with limited labeled samples. However, while existing methods have shown success in image-based tasks, their effectiveness remains unexplored for sequential time-series data from sensor systems, particularly audio inputs. To address this gap, we conduct a comprehensive benchmark study evaluating six representative Meta-CL approaches using three network architectures on five datasets from both image and audio modalities. We develop MetaCLBench, an end-to-end Meta-CL benchmark framework for edge devices to evaluate system overheads and investigate trade-offs among performance, computational costs, and memory requirements across various Meta-CL methods. Our results reveal that while many Meta-CL methods enable to learn new classes for both image and audio modalities, they impose significant computational and memory costs on edge devices. Also, we find that pre-training and meta-training procedures based on source data before deployment improve Meta-CL performance. Finally, to facilitate further research, we provide practical guidelines for researchers and machine learning practitioners implementing Meta-CL on resource-constrained environments and make our benchmark framework and tools publicly available, enabling fair evaluation across both accuracy and system-level metrics.
Continual Learning (CL) strives to learn incrementally across tasks while mitigating catastrophic forgetting. A key challenge in CL is balancing stability (retaining prior knowledge) and plasticity (learning new tasks). While representative gradient projection methods ensure stability, they often limit plasticity. Model merging techniques offer promising solutions, but prior methods typically rely on empirical assumptions and carefully selected hyperparameters. In this paper, we explore the potential of model merging to enhance the stability-plasticity trade-off, providing theoretical insights that underscore its benefits. Specifically, we reformulate the merging mechanism using Bayesian continual learning principles and derive a closed-form solution for the optimal merging coefficient that adapts to the diverse characteristics of tasks. To validate our approach, we introduce a two-stage framework named BECAME, which synergizes the expertise of gradient projection and adaptive merging. Extensive experiments show that our approach outperforms state-of-the-art CL methods and existing merging strategies.
Large Language Models (LLMs) trained on historical web data inevitably become outdated. We investigate evaluation strategies and update methods for LLMs as new data becomes available. We introduce a web-scale dataset for time-continual pretraining of LLMs derived from 114 dumps of Common Crawl (CC) - orders of magnitude larger than previous continual language modeling benchmarks. We also design time-stratified evaluations across both general CC data and specific domains (Wikipedia, StackExchange, and code documentation) to assess how well various continual learning methods adapt to new data while retaining past knowledge. Our findings demonstrate that, on general CC data, autoregressive meta-schedules combined with a fixed-ratio replay of older data can achieve comparable held-out loss to re-training from scratch, while requiring significantly less computation (2.6x). However, the optimal balance between incorporating new data and replaying old data differs as replay is crucial to avoid forgetting on generic web data but less so on specific domains.
Large Language Models (LLMs) have significantly advanced Natural Language Processing (NLP), particularly in Natural Language Understanding (NLU) tasks. As we progress toward an agentic world where LLM-based agents autonomously handle specialized tasks, it becomes crucial for these models to adapt to new tasks without forgetting previously learned information - a challenge known as catastrophic forgetting. This study evaluates the continual fine-tuning of various open-source LLMs with different parameter sizes (specifically models under 10 billion parameters) on key NLU tasks from the GLUE benchmark, including SST-2, MRPC, CoLA, and MNLI. By employing prompt engineering and task-specific adjustments, we assess and compare the models' abilities to retain prior knowledge while learning new tasks. Our results indicate that models such as Phi-3.5-mini exhibit minimal forgetting while maintaining strong learning capabilities, making them well-suited for continual learning environments. Additionally, models like Orca-2-7b and Qwen2.5-7B demonstrate impressive learning abilities and overall performance after fine-tuning. This work contributes to understanding catastrophic forgetting in LLMs and highlights prompting engineering to optimize model performance for continual learning scenarios.
Continual learning (CL) presents a fundamental challenge in training neural networks on sequential tasks without experiencing catastrophic forgetting. Traditionally, the dominant approach in CL has been gradient-based optimization, where updates to the network parameters are performed using stochastic gradient descent (SGD) or its variants. However, a major limitation arises when previous data is no longer accessible, as is often assumed in CL settings. In such cases, there is no gradient information available for past data, leading to uncontrolled parameter changes and consequently severe forgetting of previously learned tasks. By shifting focus from data availability to gradient availability, this work opens up new avenues for addressing forgetting in CL. We explore the hypothesis that gradient-free optimization methods can provide a robust alternative to conventional gradient-based continual learning approaches. We discuss the theoretical underpinnings of such method, analyze their potential advantages and limitations, and present empirical evidence supporting their effectiveness. By reconsidering the fundamental cause of forgetting, this work aims to contribute a fresh perspective to the field of continual learning and inspire novel research directions.
Humans can continuously acquire new skills and knowledge by exploiting existing ones for improved learning, without forgetting them. Similarly, 'continual learning' in machine learning aims to learn new information while preserving the previously acquired knowledge. Existing research often overlooks the nature of human learning, where tasks are interleaved due to human choice or environmental constraints. So, almost never do humans master one task before switching to the next. To investigate to what extent human-like learning can benefit the learner, we propose a method that interleaves tasks based on their 'learning progress' and energy consumption. From a machine learning perspective, our approach can be seen as a multi-task learning system that balances learning performance with energy constraints while mimicking ecologically realistic human task learning. To assess the validity of our approach, we consider a robot learning setting in simulation, where the robot learns the effect of its actions in different contexts. The conducted experiments show that our proposed method achieves better performance than sequential task learning and reduces energy consumption for learning the tasks.