Continual Learning

Deep neural networks are widely known for their remarkable effectiveness across various tasks, with the consensus that deeper networks implicitly learn more complex data representations. This paper shows that sufficiently deep networks trained for supervised image classification split into two distinct parts that contribute to the resulting data representations differently. The initial layers create linearly-separable representations, while the subsequent layers, which we refer to as \textit{the tunnel}, compress these representations and have a minimal impact on the overall performance. We explore the tunnel's behavior through comprehensive empirical studies, highlighting that it emerges early in the training process. Its depth depends on the relation between the network's capacity and task complexity. Furthermore, we show that the tunnel degrades out-of-distribution generalization and discuss its implications for continual learning.

Existing methods for adapting Neural Radiance Fields (NeRFs) to scene changes require extensive data capture and model retraining, which is both time-consuming and labor-intensive. In this paper, we tackle the challenge of efficiently adapting NeRFs to real-world scene changes over time using a few new images while retaining the memory of unaltered areas, focusing on the continual learning aspect of NeRFs. To this end, we propose CL-NeRF, which consists of two key components: a lightweight expert adaptor for adapting to new changes and evolving scene representations and a conflict-aware knowledge distillation learning objective for memorizing unchanged parts. We also present a new benchmark for evaluating Continual Learning of NeRFs with comprehensive metrics. Our extensive experiments demonstrate that CL-NeRF can synthesize high-quality novel views of both changed and unchanged regions with high training efficiency, surpassing existing methods in terms of reducing forgetting and adapting to changes. Code and benchmark will be made available.

The human brain rewires itself for neuroplasticity in the presence of new tasks. We are inspired to harness this key process in continual reinforcement learning, prioritizing adaptation to non-stationary environments. In distinction to existing rewiring approaches that rely on pruning or dynamic routing, which may limit network capacity and plasticity, this work presents a novel rewiring scheme by permuting hidden neurons. Specifically, the neuron permutation is parameterized to be end-to-end learnable and can rearrange all available synapses to explore a large span of weight space, thereby promoting adaptivity. In addition, we introduce two main designs to steer the rewiring process in continual reinforcement learning: first, a multi-mode rewiring strategy is proposed which diversifies the policy and encourages exploration when encountering new environments. Secondly, to ensure stability on history tasks, the network is devised to cache each learned wiring while subtly updating its weights, allowing for retrospective recovery of any previous state appropriate for the task. Meanwhile, an alignment mechanism is curated to achieve better plasticity-stability tradeoff by jointly optimizing cached wirings and weights. Our proposed method is comprehensively evaluated on 18 continual reinforcement learning scenarios ranging from locomotion to manipulation, demonstrating its advantages over state-of-the-art competitors in performance-efficiency tradeoffs. Code is available at https://github.com/feifeiobama/RewireNeuron.

The advancement of continual reinforcement learning (RL) has been facing various obstacles, including standardized metrics and evaluation protocols, demanding computational requirements, and a lack of widely accepted standard benchmarks. In response to these challenges, we present COOM ($\textbf{C}$ontinual D$\textbf{OOM}$), a continual RL benchmark tailored for embodied pixel-based RL. COOM presents a meticulously crafted suite of task sequences set within visually distinct 3D environments, serving as a robust evaluation framework to assess crucial aspects of continual RL, such as catastrophic forgetting, knowledge transfer, and sample-efficient learning. Following an in-depth empirical evaluation of popular continual learning (CL) methods, we pinpoint their limitations, provide valuable insight into the benchmark and highlight unique algorithmic challenges. This makes our work the first to benchmark image-based CRL in 3D environments with embodied perception. The primary objective of the COOM benchmark is to offer the research community a valuable and cost-effective challenge. It seeks to deepen our comprehension of the capabilities and limitations of current and forthcoming CL methods in an RL setting. The code and environments are open-sourced and accessible on GitHub.

Continual learning (CL) has remained a persistent challenge for deep neural networks due to catastrophic forgetting (CF) of previously learned tasks. Several techniques such as weight regularization, experience rehearsal, and parameter isolation have been proposed to alleviate CF. Despite their relative success, these research directions have predominantly remained orthogonal and suffer from several shortcomings, while missing out on the advantages of competing strategies. On the contrary, the brain continually learns, accommodates, and transfers knowledge across tasks by simultaneously leveraging several neurophysiological processes, including neurogenesis, active forgetting, neuromodulation, metaplasticity, experience rehearsal, and context-dependent gating, rarely resulting in CF. Inspired by how the brain exploits multiple mechanisms concurrently, we propose TriRE, a novel CL paradigm that encompasses retaining the most prominent neurons for each task, revising and solidifying the extracted knowledge of current and past tasks, and actively promoting less active neurons for subsequent tasks through rewinding and relearning. Across CL settings, TriRE significantly reduces task interference and surpasses different CL approaches considered in isolation.

Pretrained language models (PLMs) are today the primary model for natural language processing. Despite their impressive downstream performance, it can be difficult to apply PLMs to new languages, a barrier to making their capabilities universally accessible. While prior work has shown it possible to address this issue by learning a new embedding layer for the new language, doing so is both data and compute inefficient. We propose to use an active forgetting mechanism during pretraining, as a simple way of creating PLMs that can quickly adapt to new languages. Concretely, by resetting the embedding layer every K updates during pretraining, we encourage the PLM to improve its ability of learning new embeddings within limited number of updates, similar to a meta-learning effect. Experiments with RoBERTa show that models pretrained with our forgetting mechanism not only demonstrate faster convergence during language adaptation, but also outperform standard ones in a low-data regime, particularly for languages that are distant from English. Code will be available at https://github.com/facebookresearch/language-model-plasticity.

Continual learning (CL) aims to incrementally learn multiple tasks that are presented sequentially. The significance of CL lies not only in the practical importance but also in studying the learning mechanisms of humans who are excellent continual learners. While most research on CL has been done on structured data such as images, there is a lack of research on CL for abstract logical concepts such as counting, sorting, and arithmetic, which humans learn gradually over time in the real world. In this work, for the first time, we introduce novel algorithmic reasoning (AR) methodology for continual tasks of abstract concepts: CLeAR. Our methodology proposes a one-to-many mapping of input distribution to a shared mapping space, which allows the alignment of various tasks of different dimensions and shared semantics. Our tasks of abstract logical concepts, in the form of formal language, can be classified into Chomsky hierarchies based on their difficulty. In this study, we conducted extensive experiments consisting of 15 tasks with various levels of Chomsky hierarchy, ranging from in-hierarchy to inter-hierarchy scenarios. CLeAR not only achieved near zero forgetting but also improved accuracy during following tasks, a phenomenon known as backward transfer, while previous CL methods designed for image classification drastically failed.

Continual learning (CL) aims to train deep neural networks efficiently on streaming data while limiting the forgetting caused by new tasks. However, learning transferable knowledge with less interference between tasks is difficult, and real-world deployment of CL models is limited by their inability to measure predictive uncertainties. To address these issues, we propose handling CL tasks with neural processes (NPs), a class of meta-learners that encode different tasks into probabilistic distributions over functions all while providing reliable uncertainty estimates. Specifically, we propose an NP-based CL approach (NPCL) with task-specific modules arranged in a hierarchical latent variable model. We tailor regularizers on the learned latent distributions to alleviate forgetting. The uncertainty estimation capabilities of the NPCL can also be used to handle the task head/module inference challenge in CL. Our experiments show that the NPCL outperforms previous CL approaches. We validate the effectiveness of uncertainty estimation in the NPCL for identifying novel data and evaluating instance-level model confidence. Code is available at https://github.com/srvCodes/NPCL.

We propose and deploy an approach to continually train an instruction-following agent from feedback provided by users during collaborative interactions. During interaction, human users instruct an agent using natural language, and provide realtime binary feedback as they observe the agent following their instructions. We design a contextual bandit learning approach, converting user feedback to immediate reward. We evaluate through thousands of human-agent interactions, demonstrating 15.4% absolute improvement in instruction execution accuracy over time. We also show our approach is robust to several design variations, and that the feedback signal is roughly equivalent to the learning signal of supervised demonstration data.

Continual learning (CL) aims to incrementally learn different tasks (such as classification) in a non-stationary data stream without forgetting old ones. Most CL works focus on tackling catastrophic forgetting under a learning-from-scratch paradigm. However, with the increasing prominence of foundation models, pre-trained models equipped with informative representations have become available for various downstream requirements. Several CL methods based on pre-trained models have been explored, either utilizing pre-extracted features directly (which makes bridging distribution gaps challenging) or incorporating adaptors (which may be subject to forgetting). In this paper, we propose a concise and effective approach for CL with pre-trained models. Given that forgetting occurs during parameter updating, we contemplate an alternative approach that exploits training-free random projectors and class-prototype accumulation, which thus bypasses the issue. Specifically, we inject a frozen Random Projection layer with nonlinear activation between the pre-trained model's feature representations and output head, which captures interactions between features with expanded dimensionality, providing enhanced linear separability for class-prototype-based CL. We also demonstrate the importance of decorrelating the class-prototypes to reduce the distribution disparity when using pre-trained representations. These techniques prove to be effective and circumvent the problem of forgetting for both class- and domain-incremental continual learning. Compared to previous methods applied to pre-trained ViT-B/16 models, we reduce final error rates by between 20% and 62% on seven class-incremental benchmark datasets, despite not using any rehearsal memory. We conclude that the full potential of pre-trained models for simple, effective, and fast continual learning has not hitherto been fully tapped. Code is available at https://github.com/RanPAC/RanPAC.

In a standard view of the reinforcement learning problem, an agent’s goal is to efficiently identify a policy that maximizes long-term reward. However, this perspective is based on a restricted view of learning as finding a solution, rather than treating learning as endless adaptation. In contrast, continual reinforcement learning refers to the setting in which the best agents never stop learning. Despite the importance of continual reinforcement learning, the community lacks a simple definition of the problem that highlights its commitments and makes its primary concepts precise and clear. To this end, this paper is dedicated to carefully defining the continual reinforcement learning problem. We formalize the notion of agents that “never stop learning” through a new mathematical language for analyzing and cataloging agents. Using this new language, we define a continual learning agent as one that can be understood as carrying out an implicit search process indefinitely, and continual reinforcement learning as the setting in which the best agents are all continual learning agents. We provide two motivating examples, illustrating that traditional views of multi-task reinforcement learning and continual supervised learning are special cases of our definition. Collectively, these definitions and perspectives formalize many intuitive concepts at the heart of learning, and open new research pathways surrounding continual learning agents.

We provide theoretical bounds on the forgetting quantity in the continual learning setting for linear tasks, where each round of learning corresponds to projecting onto a linear subspace. For a cyclic task ordering on $T$ tasks repeated $m$ times each, we prove the best known upper bound of $O(T^2/m)$ on the forgetting. Notably, our bound holds uniformly over all choices of tasks and is independent of the ambient dimension. Our main technical contribution is a characterization of the union of all numerical ranges of products of $T$ (real or complex) projections as a sinusoidal spiral, which may be of independent interest.

Test-Time Adaptation (TTA) allows to update pre-trained models to changing data distributions at deployment time. While early work tested these algorithms for individual fixed distribution shifts, recent work proposed and applied methods for continual adaptation over long timescales. To examine the reported progress in the field, we propose the Continually Changing Corruptions (CCC) benchmark to measure asymptotic performance of TTA techniques. We find that eventually all but one state-of-the-art methods collapse and perform worse than a non-adapting model, including models specifically proposed to be robust to performance collapse. In addition, we introduce a simple baseline, "RDumb", that periodically resets the model to its pretrained state. RDumb performs better or on par with the previously proposed state-of-the-art in all considered benchmarks. Our results show that previous TTA approaches are neither effective at regularizing adaptation to avoid collapse nor able to outperform a simplistic resetting strategy.

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.

Deployed language models decay over time due to shifting inputs, changing user needs, or emergent world-knowledge gaps. When such problems are identified, we want to make targeted edits while avoiding expensive retraining. However, current model editors, which modify such behaviors of pre-trained models, degrade model performance quickly across multiple, sequential edits. We propose GRACE, a \textit{lifelong} model editing method, which implements spot-fixes on streaming errors of a deployed model, ensuring minimal impact on unrelated inputs. GRACE writes new mappings into a pre-trained model's latent space, creating a discrete, local codebook of edits without altering model weights. This is the first method enabling thousands of sequential edits using only streaming errors. Our experiments on T5, BERT, and GPT models show GRACE's state-of-the-art performance in making and retaining edits, while generalizing to unseen inputs. Our code is available at [github.com/thartvigsen/grace](https://www.github.com/thartvigsen/grace}).

Dataset condensation (DC) distills a large real-world dataset into a small synthetic dataset, with the goal of training a network from scratch on the latter that performs similarly to the former. State-of-the-art (SOTA) DC methods have achieved satisfactory results through techniques such as accuracy, gradient, training trajectory, or distribution matching. However, these works all perform matching in the high-dimension pixel spaces, ignoring that natural images are usually locally connected and have lower intrinsic dimensions, resulting in low condensation efficiency. In this work, we propose a simple-yet-efficient dataset condensation plugin that matches the raw and synthetic datasets in a low-dimensional manifold. Specifically, our plugin condenses raw images into two low-rank matrices instead of parameterized image matrices. Our plugin can be easily incorporated into existing DC methods, thereby containing richer raw dataset information at limited storage costs to improve the downstream applications' performance. We verify on multiple public datasets that when the proposed plugin is combined with SOTA DC methods, the performance of the network trained on synthetic data is significantly improved compared to traditional DC methods. Moreover, when applying the DC methods as a plugin to continual learning tasks, we observed that our approach effectively mitigates catastrophic forgetting of old tasks under limited memory buffer constraints and avoids the problem of raw data privacy leakage.

Neural Temporal Point Processes (TPPs) are the prevalent paradigm for modeling continuous-time event sequences, such as user activities on the web and financial transactions. In real world applications, the event data typically comes in a streaming manner, where the distribution of the patterns may shift over time. Under the privacy and memory constraints commonly seen in real scenarios, how to continuously monitor a TPP to learn the streaming event sequence is an important yet under-investigated problem. In this work, we approach this problem by adopting Continual Learning (CL), which aims to enable a model to continuously learn a sequence of tasks without catastrophic forgetting. While CL for event sequence is less well studied, we present a simple yet effective framework, PromptTPP, by integrating the base TPP with a continuous-time retrieval prompt pool. In our proposed framework, prompts are small learnable parameters, maintained in a memory space and jointly optimized with the base TPP so that the model is properly instructed to learn event streams arriving sequentially without buffering past examples or task-specific attributes. We formalize a novel and realistic experimental setup for modeling event streams, where PromptTPP consistently sets state-of-the-art performance across two real user behavior datasets.

Continual semantic segmentation aims to learn new classes while maintaining the information from the previous classes. Although prior studies have shown impressive progress in recent years, the fairness concern in the continual semantic segmentation needs to be better addressed. Meanwhile, fairness is one of the most vital factors in deploying the deep learning model, especially in human-related or safety applications. In this paper, we present a novel Fairness Continual Learning approach to the semantic segmentation problem. In particular, under the fairness objective, a new fairness continual learning framework is proposed based on class distributions. Then, a novel Prototypical Contrastive Clustering loss is proposed to address the significant challenges in continual learning, i.e., catastrophic forgetting and background shift. Our proposed loss has also been proven as a novel, generalized learning paradigm of knowledge distillation commonly used in continual learning. Moreover, the proposed Conditional Structural Consistency loss further regularized the structural constraint of the predicted segmentation. Our proposed approach has achieved State-of-the-Art performance on three standard scene understanding benchmarks, i.e., ADE20K, Cityscapes, and Pascal VOC, and promoted the fairness of the segmentation model.

We focus on Continual Meta-Learning (CML), which targets accumulating and exploiting meta-knowledge on a sequence of non-i.i.d. tasks. The primary challenge is to strike a balance between stability and plasticity, where a model should be stable to avoid catastrophic forgetting in previous tasks and plastic to learn generalizable concepts from new tasks. To address this, we formulate the CML objective as controlling the average excess risk upper bound of the task sequence, which reflects the trade-off between forgetting and generalization. Based on the objective, we introduce a unified theoretical framework for CML in both static and shifting environments, providing guarantees for various task-specific learning algorithms. Moreover, we first present a rigorous analysis of a bi-level trade-off in shifting environments. To approach the optimal trade-off, we propose a novel algorithm that dynamically adjusts the meta-parameter and its learning rate w.r.t environment change. Empirical evaluations on synthetic and real datasets illustrate the effectiveness of the proposed theory and algorithm.

Existing class-incremental semantic segmentation (CISS) methods mainly tackle catastrophic forgetting and background shift, but often overlook another crucial issue. In CISS, each step focuses on different foreground classes, and the training set for a single step only includes images containing pixels of the current foreground classes, excluding images without them. This leads to an overrepresentation of these foreground classes in the single-step training set, causing the classification biased towards these classes. To address this issue, we present STAR, which preserves the main characteristics of each past class by storing a compact prototype and necessary statistical data, and aligns the class distribution of single-step training samples with the complete dataset by replaying these prototypes and repeating background pixels with appropriate frequency. Compared to the previous works that replay raw images, our method saves over 100 times the storage while achieving better performance. Moreover, STAR incorporates an old-class features maintaining (OCFM) loss, keeping old-class features unchanged while preserving sufficient plasticity for learning new classes. Furthermore, a similarity-aware discriminative (SAD) loss is employed to specifically enhance the feature diversity between similar old-new class pairs. Experiments on two public datasets, Pascal VOC 2012 and ADE20K, reveal that our model surpasses all previous state-of-the-art methods.