Artificial neural networks, celebrated for their human-like cognitive learning abilities, often encounter the well-known catastrophic forgetting (CF) problem, where the neural networks lose the proficiency in previously acquired knowledge. Despite numerous efforts to mitigate CF, it remains the significant challenge particularly in complex changing environments. This challenge is even more pronounced in cross-domain adaptation following the continual learning (CL) setting, which is a more challenging and realistic scenario that is under-explored. To this end, this article proposes a cross-domain CL approach making possible to deploy a single model in such environments without additional labelling costs. Our approach, namely continual learning approach for many processes (CLAMP), integrates a class-aware adversarial domain adaptation strategy to align a source domain and a target domain. An assessor-guided learning process is put forward to navigate the learning process of a base model assigning a set of weights to every sample controlling the influence of every sample and the interactions of each loss function in such a way to balance the stability and plasticity dilemma thus preventing the CF problem. The first assessor focuses on the negative transfer problem rejecting irrelevant samples of the source domain while the second assessor prevents noisy pseudo labels of the target domain. Both assessors are trained in the meta-learning approach using random transformation techniques and similar samples of the source domain. Theoretical analysis and extensive numerical validations demonstrate that CLAMP significantly outperforms established baseline algorithms across all experiments by at least $10\%$ margin.
Aspect-based sentiment analysis (ABSA) is an important subtask of sentiment analysis, which aims to extract the aspects and predict their sentiments. Most existing studies focus on improving the performance of the target domain by fine-tuning domain-specific models (trained on source domains) based on the target domain dataset. Few works propose continual learning tasks for ABSA, which aim to learn the target domain's ability while maintaining the history domains' abilities. In this paper, we propose a Large Language Model-based Continual Learning (\texttt{LLM-CL}) model for ABSA. First, we design a domain knowledge decoupling module to learn a domain-invariant adapter and separate domain-variant adapters dependently with an orthogonal constraint. Then, we introduce a domain knowledge warmup strategy to align the representation between domain-invariant and domain-variant knowledge. In the test phase, we index the corresponding domain-variant knowledge via domain positioning to not require each sample's domain ID. Extensive experiments over 19 datasets indicate that our \texttt{LLM-CL} model obtains new state-of-the-art performance.
Continual learning (CL) provides a framework for training models in ever-evolving environments. Although re-occurrence of previously seen objects or tasks is common in real-world problems, the concept of repetition in the data stream is not often considered in standard benchmarks for CL. Unlike with the rehearsal mechanism in buffer-based strategies, where sample repetition is controlled by the strategy, repetition in the data stream naturally stems from the environment. This report provides a summary of the CLVision challenge at CVPR 2023, which focused on the topic of repetition in class-incremental learning. The report initially outlines the challenge objective and then describes three solutions proposed by finalist teams that aim to effectively exploit the repetition in the stream to learn continually. The experimental results from the challenge highlight the effectiveness of ensemble-based solutions that employ multiple versions of similar modules, each trained on different but overlapping subsets of classes. This report underscores the transformative potential of taking a different perspective in CL by employing repetition in the data stream to foster innovative strategy design.
Continual learning (CL) has spurred the development of several methods aimed at consolidating previous knowledge across sequential learning. Yet, the evaluations of these methods have primarily focused on the final output, such as changes in the accuracy of predicted classes, overlooking the issue of representational forgetting within the model. In this paper, we propose a novel representation-based evaluation framework for CL models. This approach involves gathering internal representations from throughout the continual learning process and formulating three-dimensional tensors. The tensors are formed by stacking representations, such as layer activations, generated from several inputs and model `snapshots', throughout the learning process. By conducting tensor component analysis (TCA), we aim to uncover meaningful patterns about how the internal representations evolve, expecting to highlight the merits or shortcomings of examined CL strategies. We conduct our analyses across different model architectures and importance-based continual learning strategies, with a curated task selection. While the results of our approach mirror the difference in performance of various CL strategies, we found that our methodology did not directly highlight specialized clusters of neurons, nor provide an immediate understanding the evolution of filters. We believe a scaled down version of our approach will provide insight into the benefits and pitfalls of using TCA to study continual learning dynamics.
While humans excel at continual learning (CL), deep neural networks (DNNs) exhibit catastrophic forgetting. A salient feature of the brain that allows effective CL is that it utilizes multiple modalities for learning and inference, which is underexplored in DNNs. Therefore, we study the role and interactions of multiple modalities in mitigating forgetting and introduce a benchmark for multimodal continual learning. Our findings demonstrate that leveraging multiple views and complementary information from multiple modalities enables the model to learn more accurate and robust representations. This makes the model less vulnerable to modality-specific regularities and considerably mitigates forgetting. Furthermore, we observe that individual modalities exhibit varying degrees of robustness to distribution shift. Finally, we propose a method for integrating and aligning the information from different modalities by utilizing the relational structural similarities between the data points in each modality. Our method sets a strong baseline that enables both single- and multimodal inference. Our study provides a promising case for further exploring the role of multiple modalities in enabling CL and provides a standard benchmark for future research.
Web phishing poses a dynamic threat, requiring detection systems to quickly adapt to the latest tactics. Traditional approaches of accumulating data and periodically retraining models are outpaced. We propose a novel paradigm combining federated learning and continual learning, enabling distributed nodes to continually update models on streams of new phishing data, without accumulating data. These locally adapted models are then aggregated at a central server via federated learning. To enhance detection, we introduce a custom attention-based classifier model with residual connections, tailored for web phishing, leveraging attention mechanisms to capture intricate phishing patterns. We evaluate our hybrid learning paradigm across continual learning strategies (cumulative, replay, MIR, LwF) and model architectures through an empirical investigation. Our main contributions are: (1) a new hybrid federated-continual learning paradigm for robust web phishing detection, and (2) a novel attention + residual connections based model explicitly designed for this task, attaining 0.93 accuracy, 0.90 precision, 0.96 recall and 0.93 f1-score with the LwF strategy, outperforming traditional approaches in detecting emerging phishing threats while retaining past knowledge.
Continual learning entails learning a sequence of tasks and balancing their knowledge appropriately. With limited access to old training samples, much of the current work in deep neural networks has focused on overcoming catastrophic forgetting of old tasks in gradient-based optimization. However, the normalization layers provide an exception, as they are updated interdependently by the gradient and statistics of currently observed training samples, which require specialized strategies to mitigate recency bias. In this work, we focus on the most popular Batch Normalization (BN) and provide an in-depth theoretical analysis of its sub-optimality in continual learning. Our analysis demonstrates the dilemma between balance and adaptation of BN statistics for incremental tasks, which potentially affects training stability and generalization. Targeting on these particular challenges, we propose Adaptive Balance of BN (AdaB$^2$N), which incorporates appropriately a Bayesian-based strategy to adapt task-wise contributions and a modified momentum to balance BN statistics, corresponding to the training and testing stages. By implementing BN in a continual learning fashion, our approach achieves significant performance gains across a wide range of benchmarks, particularly for the challenging yet realistic online scenarios (e.g., up to 7.68%, 6.86% and 4.26% on Split CIFAR-10, Split CIFAR-100 and Split Mini-ImageNet, respectively). Our code is available at https://github.com/lvyilin/AdaB2N.
Lower limb amputations and neuromuscular impairments severely restrict mobility, necessitating advancements beyond conventional prosthetics. While motorized bionic limbs show promise, their effectiveness depends on replicating the dynamic coordination of human movement across diverse environments. In this paper, we introduce a model for human behavior in the context of bionic prosthesis control. Our approach leverages human locomotion demonstrations to learn the synergistic coupling of the lower limbs, enabling the prediction of the kinematic behavior of a missing limb during tasks such as walking, climbing inclines, and stairs. We propose a multitasking, continually adaptive model that anticipates and refines movements over time. At the core of our method is a technique called multitask prospective rehearsal, that anticipates and synthesizes future movements based on the previous prediction and employs a corrective mechanism for subsequent predictions. Our evolving architecture merges lightweight, task-specific modules on a shared backbone, ensuring both specificity and scalability. We validate our model through experiments on real-world human gait datasets, including transtibial amputees, across a wide range of locomotion tasks. Results demonstrate that our approach consistently outperforms baseline models, particularly in scenarios with distributional shifts, adversarial perturbations, and noise.
In autonomous and mobile robotics, a principal challenge is resilient real-time environmental perception, particularly in situations characterized by unknown and dynamic elements, as exemplified in the context of autonomous drone racing. This study introduces a perception technique for detecting drone racing gates under illumination variations, which is common during high-speed drone flights. The proposed technique relies upon a lightweight neural network backbone augmented with capabilities for continual learning. The envisaged approach amalgamates predictions of the gates' positional coordinates, distance, and orientation, encapsulating them into a cohesive pose tuple. A comprehensive number of tests serve to underscore the efficacy of this approach in confronting diverse and challenging scenarios, specifically those involving variable lighting conditions. The proposed methodology exhibits notable robustness in the face of illumination variations, thereby substantiating its effectiveness.
While the human brain efficiently adapts to new tasks from a continuous stream of information, neural network models struggle to learn from sequential information without catastrophically forgetting previously learned tasks. This limitation presents a significant hurdle in deploying edge devices in real-world scenarios where information is presented in an inherently sequential manner. Active dendrites of pyramidal neurons play an important role in the brain ability to learn new tasks incrementally. By exploiting key properties of time-to-first-spike encoding and leveraging its high sparsity, we present a novel spiking neural network model enhanced with active dendrites. Our model can efficiently mitigate catastrophic forgetting in temporally-encoded SNNs, which we demonstrate with an end-of-training accuracy across tasks of 88.3% on the test set using the Split MNIST dataset. Furthermore, we provide a novel digital hardware architecture that paves the way for real-world deployment in edge devices. Using a Xilinx Zynq-7020 SoC FPGA, we demonstrate a 100-% match with our quantized software model, achieving an average inference time of 37.3 ms and an 80.0% accuracy.
We formulate a unifying framework for unsupervised continual learning (UCL), which disentangles learning objectives that are specific to the present and the past data, encompassing stability, plasticity, and cross-task consolidation. The framework reveals that many existing UCL approaches overlook cross-task consolidation and try to balance plasticity and stability in a shared embedding space. This results in worse performance due to a lack of within-task data diversity and reduced effectiveness in learning the current task. Our method, Osiris, which explicitly optimizes all three objectives on separate embedding spaces, achieves state-of-the-art performance on all benchmarks, including two novel benchmarks proposed in this paper featuring semantically structured task sequences. Compared to standard benchmarks, these two structured benchmarks more closely resemble visual signals received by humans and animals when navigating real-world environments. Finally, we show some preliminary evidence that continual models can benefit from such realistic learning scenarios.
Large language models (LLMs) with one or more fine-tuning phases have become a necessary step to unlock various capabilities, enabling LLMs to follow natural language instructions or align with human preferences. However, it carries the risk of catastrophic forgetting during sequential training, the parametric knowledge or the ability learned in previous stages may be overwhelmed by incoming training data. In this paper, we find that by regularly resetting partial parameters, LLMs can restore some of the original knowledge. Inspired by this, we introduce Half Fine-Tuning (HFT) for LLMs, as a substitute for full fine-tuning (FFT), to mitigate the forgetting issues, where half of the parameters are selected to learn new tasks while the other half are frozen to remain previous knowledge. We provide a feasibility analysis from the perspective of optimization and interpret the parameter selection operation as a regularization term. Without changing the model architecture, HFT could be seamlessly integrated into existing fine-tuning frameworks. Extensive experiments and analysis on supervised fine-tuning, direct preference optimization, and continual learning consistently demonstrate the effectiveness, robustness, and efficiency of HFT. Compared with FFT, HFT not only significantly alleviates the forgetting problem, but also achieves the best performance in a series of downstream benchmarks, with an approximately 30% reduction in training time.
Contemporary continual learning approaches typically select prompts from a pool, which function as supplementary inputs to a pre-trained model. However, this strategy is hindered by the inherent noise of its selection approach when handling increasing tasks. In response to these challenges, we reformulate the prompting approach for continual learning and propose the prompt customization (PC) method. PC mainly comprises a prompt generation module (PGM) and a prompt modulation module (PMM). In contrast to conventional methods that employ hard prompt selection, PGM assigns different coefficients to prompts from a fixed-sized pool of prompts and generates tailored prompts. Moreover, PMM further modulates the prompts by adaptively assigning weights according to the correlations between input data and corresponding prompts. We evaluate our method on four benchmark datasets for three diverse settings, including the class, domain, and task-agnostic incremental learning tasks. Experimental results demonstrate consistent improvement (by up to 16.2\%), yielded by the proposed method, over the state-of-the-art (SOTA) techniques.
Incremental learning is the ability of systems to acquire knowledge over time, enabling their adaptation and generalization to novel tasks. It is a critical ability for intelligent, real-world systems, especially when data changes frequently or is limited. This review provides a comprehensive analysis of incremental learning in Large Language Models. It synthesizes the state-of-the-art incremental learning paradigms, including continual learning, meta-learning, parameter-efficient learning, and mixture-of-experts learning. We demonstrate their utility for incremental learning by describing specific achievements from these related topics and their critical factors. An important finding is that many of these approaches do not update the core model, and none of them update incrementally in real-time. The paper highlights current problems and challenges for future research in the field. By consolidating the latest relevant research developments, this review offers a comprehensive understanding of incremental learning and its implications for designing and developing LLM-based learning systems.
Continual learning (CL) remains one of the long-standing challenges for deep neural networks due to catastrophic forgetting of previously acquired knowledge. Although rehearsal-based approaches have been fairly successful in mitigating catastrophic forgetting, they suffer from overfitting on buffered samples and prior information loss, hindering generalization under low-buffer regimes. Inspired by how humans learn using strong inductive biases, we propose IMEX-Reg to improve the generalization performance of experience rehearsal in CL under low buffer regimes. Specifically, we employ a two-pronged implicit-explicit regularization approach using contrastive representation learning (CRL) and consistency regularization. To further leverage the global relationship between representations learned using CRL, we propose a regularization strategy to guide the classifier toward the activation correlations in the unit hypersphere of the CRL. Our results show that IMEX-Reg significantly improves generalization performance and outperforms rehearsal-based approaches in several CL scenarios. It is also robust to natural and adversarial corruptions with less task-recency bias. Additionally, we provide theoretical insights to support our design decisions further.
Artificial neural networks often suffer from catastrophic forgetting, where learning new concepts leads to a complete loss of previously acquired knowledge. We observe that this issue is particularly magnified in vision transformers (ViTs), where post-pre-training and fine-tuning on new tasks can significantly degrade the model's original general abilities. For instance, a DINO ViT-Base/16 pre-trained on ImageNet-1k loses over 70% accuracy on ImageNet-1k after just 10 iterations of fine-tuning on CIFAR-100. Overcoming this stability-plasticity dilemma is crucial for enabling ViTs to continuously learn and adapt to new domains while preserving their initial knowledge. In this work, we study two new parameter-efficient fine-tuning strategies: (1)~Block Expansion, and (2) Low-rank adaptation (LoRA). Our experiments reveal that using either Block Expansion or LoRA on self-supervised pre-trained ViTs surpass fully fine-tuned ViTs in new domains while offering significantly greater parameter efficiency. Notably, we find that Block Expansion experiences only a minimal performance drop in the pre-training domain, thereby effectively mitigating catastrophic forgetting in pre-trained ViTs.
The recent success of large language models (LLMs) trained on static, pre-collected, general datasets has sparked numerous research directions and applications. One such direction addresses the non-trivial challenge of integrating pre-trained LLMs into dynamic data distributions, task structures, and user preferences. Pre-trained LLMs, when tailored for specific needs, often experience significant performance degradation in previous knowledge domains -- a phenomenon known as "catastrophic forgetting". While extensively studied in the continual learning (CL) community, it presents new manifestations in the realm of LLMs. In this survey, we provide a comprehensive overview of the current research progress on LLMs within the context of CL. This survey is structured into four main sections: we first describe an overview of continually learning LLMs, consisting of two directions of continuity: vertical continuity (or vertical continual learning), i.e., continual adaptation from general to specific capabilities, and horizontal continuity (or horizontal continual learning), i.e., continual adaptation across time and domains (Section 3). We then summarize three stages of learning LLMs in the context of modern CL: Continual Pre-Training (CPT), Domain-Adaptive Pre-training (DAP), and Continual Fine-Tuning (CFT) (Section 4). Then we provide an overview of evaluation protocols for continual learning with LLMs, along with the current available data sources (Section 5). Finally, we discuss intriguing questions pertaining to continual learning for LLMs (Section 6). The full list of papers examined in this survey is available at https://github.com/Wang-ML-Lab/llm-continual-learning-survey.
Modeling non-stationary data is a challenging problem in the field of continual learning, and data distribution shifts may result in negative consequences on the performance of a machine learning model. Classic learning tools are often vulnerable to perturbations of the input covariates, and are sensitive to outliers and noise, and some tools are based on rigid algebraic assumptions. Distribution shifts are frequently occurring due to changes in raw materials for production, seasonality, a different user base, or even adversarial attacks. Therefore, there is a need for more effective distribution shift detection techniques. In this work, we propose a continual learning framework for monitoring and detecting distribution changes. We explore the problem in a latent space generated by a bio-inspired self-organizing clustering and statistical aspects of the latent space. In particular, we investigate the projections made by two topology-preserving maps: the Self-Organizing Map and the Scale Invariant Map. Our method can be applied in both a supervised and an unsupervised context. We construct the assessment of changes in the data distribution as a comparison of Gaussian signals, making the proposed method fast and robust. We compare it to other unsupervised techniques, specifically Principal Component Analysis (PCA) and Kernel-PCA. Our comparison involves conducting experiments using sequences of images (based on MNIST and injected shifts with adversarial samples), chemical sensor measurements, and the environmental variable related to ozone levels. The empirical study reveals the potential of the proposed approach.
Artificial neural networks have advanced due to scaling dimensions, but conventional computing faces inefficiency due to the von Neumann bottleneck. In-memory computation architectures, like memristors, offer promise but face challenges due to hardware non-idealities. This work proposes and experimentally demonstrates layer ensemble averaging, a technique to map pre-trained neural network solutions from software to defective hardware crossbars of emerging memory devices and reliably attain near-software performance on inference. The approach is investigated using a custom 20,000-device hardware prototyping platform on a continual learning problem where a network must learn new tasks without catastrophically forgetting previously learned information. Results demonstrate that by trading off the number of devices required for layer mapping, layer ensemble averaging can reliably boost defective memristive network performance up to the software baseline. For the investigated problem, the average multi-task classification accuracy improves from 61 % to 72 % (< 1 % of software baseline) using the proposed approach.
Efforts to overcome catastrophic forgetting have primarily centered around developing more effective Continual Learning (CL) methods. In contrast, less attention was devoted to analyzing the role of network architecture design (e.g., network depth, width, and components) in contributing to CL. This paper seeks to bridge this gap between network architecture design and CL, and to present a holistic study on the impact of network architectures on CL. This work considers architecture design at the network scaling level, i.e., width and depth, and also at the network components, i.e., skip connections, global pooling layers, and down-sampling. In both cases, we first derive insights through systematically exploring how architectural designs affect CL. Then, grounded in these insights, we craft a specialized search space for CL and further propose a simple yet effective ArchCraft method to steer a CL-friendly architecture, namely, this method recrafts AlexNet/ResNet into AlexAC/ResAC. Experimental validation across various CL settings and scenarios demonstrates that improved architectures are parameter-efficient, achieving state-of-the-art performance of CL while being 86%, 61%, and 97% more compact in terms of parameters than the naive CL architecture in Task IL and Class IL. Code is available at https://github.com/byyx666/ArchCraft.