PyTorchKR Papers

최신 논문

총 2,786편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2309.06896

Domain-Aware Augmentations for Unsupervised Online General Continual Learning

Continual Learning has been challenging, especially when dealing with unsupervised scenarios such as Unsupervised Online General Continual Learning (UOGCL), where the learning agent has no prior knowledge of class boundaries or task change information. While previous research has focused on reducing forgetting in supervised setups, recent studies have shown that self-supervised learners are more resilient to forgetting. This paper proposes a novel approach that enhances memory usage for contrastive learning in UOGCL by defining and using stream-dependent data augmentations together with some implementation tricks. Our proposed method is simple yet effective, achieves state-of-the-art results compared to other unsupervised approaches in all considered setups, and reduces the gap between supervised and unsupervised continual learning. Our domain-aware augmentation procedure can be adapted to other replay-based methods, making it a promising strategy for continual learning.

arXiv:2309.06917

Continual Learning with Dirichlet Generative-based Rehearsal

Recent advancements in data-driven task-oriented dialogue systems (ToDs) struggle with incremental learning due to computational constraints and time-consuming issues. Continual Learning (CL) attempts to solve this by avoiding intensive pre-training, but it faces the problem of catastrophic forgetting (CF). While generative-based rehearsal CL methods have made significant strides, generating pseudo samples that accurately reflect the underlying task-specific distribution is still a challenge. In this paper, we present Dirichlet Continual Learning (DCL), a novel generative-based rehearsal strategy for CL. Unlike the traditionally used Gaussian latent variable in the Conditional Variational Autoencoder (CVAE), DCL leverages the flexibility and versatility of the Dirichlet distribution to model the latent prior variable. This enables it to efficiently capture sentence-level features of previous tasks and effectively guide the generation of pseudo samples. In addition, we introduce Jensen-Shannon Knowledge Distillation (JSKD), a robust logit-based knowledge distillation method that enhances knowledge transfer during pseudo sample generation. Our experiments confirm the efficacy of our approach in both intent detection and slot-filling tasks, outperforming state-of-the-art methods.

arXiv:2309.06256공식 코드 · Papers with Code

Mitigating the Alignment Tax of RLHF

LLMs acquire a wide range of abilities during pre-training, but aligning LLMs under Reinforcement Learning with Human Feedback (RLHF) can lead to forgetting pretrained abilities, which is also known as the alignment tax. To investigate alignment tax, we conducted experiments with existing RLHF algorithms using OpenLLaMA-3B, which revealed a pronounced alignment tax in NLP tasks. Whereas, despite various techniques to mitigate forgetting, they are often at odds with the RLHF performance, leading to a trade-off between alignment performance and forgetting mitigation, leading to an alignment-forgetting trade-off. In this paper we show that model averaging, which simply interpolates between pre and post RLHF model weights, surprisingly achieves the most strongest alignment-forgetting Pareto front among a wide range of competing methods. To understand its effectiveness, we offer theoretical insights into model averaging, revealing that it enhances performance Pareto front by increasing feature diversity on the layers where tasks share overlapped feature spaces. Empirical evidence corroborates our analysis by showing the benefits of averaging low-level transformer layers. Building on the analysis and the observation that averaging different layers of the transformer leads to significantly different alignment-forgetting trade-offs, we propose Heterogeneous Model Averaging (HMA) to Heterogeneously find various combination ratios of model layers. HMA seeks to maximize the alignment performance while incurring minimal alignment tax. Moreover, we validate HMA's performance across a range of RLHF algorithms over OpenLLaMA-3B and further extend our findings to Mistral-7B which is evaluated by open-sourced preference model and GPT4. Code available here: https://github.com/avalonstrel/Mitigating-the-Alignment-Tax-of-RLHF.git.

arXiv:2309.06086공식 코드 · Papers with Code

Plasticity-Optimized Complementary Networks for Unsupervised Continual Learning

Continuous unsupervised representation learning (CURL) research has greatly benefited from improvements in self-supervised learning (SSL) techniques. As a result, existing CURL methods using SSL can learn high-quality representations without any labels, but with a notable performance drop when learning on a many-tasks data stream. We hypothesize that this is caused by the regularization losses that are imposed to prevent forgetting, leading to a suboptimal plasticity-stability trade-off: they either do not adapt fully to the incoming data (low plasticity), or incur significant forgetting when allowed to fully adapt to a new SSL pretext-task (low stability). In this work, we propose to train an expert network that is relieved of the duty of keeping the previous knowledge and can focus on performing optimally on the new tasks (optimizing plasticity). In the second phase, we combine this new knowledge with the previous network in an adaptation-retrospection phase to avoid forgetting and initialize a new expert with the knowledge of the old network. We perform several experiments showing that our proposed approach outperforms other CURL exemplar-free methods in few- and many-task split settings. Furthermore, we show how to adapt our approach to semi-supervised continual learning (Semi-SCL) and show that we surpass the accuracy of other exemplar-free Semi-SCL methods and reach the results of some others that use exemplars.

arXiv:2309.05281공식 코드 · Papers with Code

Class-Incremental Grouping Network for Continual Audio-Visual Learning

Continual learning is a challenging problem in which models need to be trained on non-stationary data across sequential tasks for class-incremental learning. While previous methods have focused on using either regularization or rehearsal-based frameworks to alleviate catastrophic forgetting in image classification, they are limited to a single modality and cannot learn compact class-aware cross-modal representations for continual audio-visual learning. To address this gap, we propose a novel class-incremental grouping network (CIGN) that can learn category-wise semantic features to achieve continual audio-visual learning. Our CIGN leverages learnable audio-visual class tokens and audio-visual grouping to continually aggregate class-aware features. Additionally, it utilizes class tokens distillation and continual grouping to prevent forgetting parameters learned from previous tasks, thereby improving the model's ability to capture discriminative audio-visual categories. We conduct extensive experiments on VGGSound-Instruments, VGGSound-100, and VGG-Sound Sources benchmarks. Our experimental results demonstrate that the CIGN achieves state-of-the-art audio-visual class-incremental learning performance. Code is available at https://github.com/stoneMo/CIGN.

arXiv:2309.04974

Continual Robot Learning using Self-Supervised Task Inference

Endowing robots with the human ability to learn a growing set of skills over the course of a lifetime as opposed to mastering single tasks is an open problem in robot learning. While multi-task learning approaches have been proposed to address this problem, they pay little attention to task inference. In order to continually learn new tasks, the robot first needs to infer the task at hand without requiring predefined task representations. In this paper, we propose a self-supervised task inference approach. Our approach learns action and intention embeddings from self-organization of the observed movement and effect parts of unlabeled demonstrations and a higher-level behavior embedding from self-organization of the joint action-intention embeddings. We construct a behavior-matching self-supervised learning objective to train a novel Task Inference Network (TINet) to map an unlabeled demonstration to its nearest behavior embedding, which we use as the task representation. A multi-task policy is built on top of the TINet and trained with reinforcement learning to optimize performance over tasks. We evaluate our approach in the fixed-set and continual multi-task learning settings with a humanoid robot and compare it to different multi-task learning baselines. The results show that our approach outperforms the other baselines, with the difference being more pronounced in the challenging continual learning setting, and can infer tasks from incomplete demonstrations. Our approach is also shown to generalize to unseen tasks based on a single demonstration in one-shot task generalization experiments.

arXiv:2309.02210

Continual Cross-Dataset Adaptation in Road Surface Classification

Accurate road surface classification is crucial for autonomous vehicles (AVs) to optimize driving conditions, enhance safety, and enable advanced road mapping. However, deep learning models for road surface classification suffer from poor generalization when tested on unseen datasets. To update these models with new information, also the original training dataset must be taken into account, in order to avoid catastrophic forgetting. This is, however, inefficient if not impossible, e.g., when the data is collected in streams or large amounts. To overcome this limitation and enable fast and efficient cross-dataset adaptation, we propose to employ continual learning finetuning methods designed to retain past knowledge while adapting to new data, thus effectively avoiding forgetting. Experimental results demonstrate the superiority of this approach over naive finetuning, achieving performance close to fresh retraining. While solving this known problem, we also provide a general description of how the same technique can be adopted in other AV scenarios. We highlight the potential computational and economic benefits that a continual-based adaptation can bring to the AV industry, while also reducing greenhouse emissions due to unnecessary joint retraining.

arXiv:2309.01612

On the Query Strategies for Efficient Online Active Distillation

Deep Learning (DL) requires lots of time and data, resulting in high computational demands. Recently, researchers employ Active Learning (AL) and online distillation to enhance training efficiency and real-time model adaptation. This paper evaluates a set of query strategies to achieve the best training results. It focuses on Human Pose Estimation (HPE) applications, assessing the impact of selected frames during training using two approaches: a classical offline method and a online evaluation through a continual learning approach employing knowledge distillation, on a popular state-of-the-art HPE dataset. The paper demonstrates the possibility of enabling training at the edge lightweight models, adapting them effectively to new contexts in real-time.

arXiv:2309.02995공식 코드 · Papers with Code

Continual Evidential Deep Learning for Out-of-Distribution Detection

Uncertainty-based deep learning models have attracted a great deal of interest for their ability to provide accurate and reliable predictions. Evidential deep learning stands out achieving remarkable performance in detecting out-of-distribution (OOD) data with a single deterministic neural network. Motivated by this fact, in this paper we propose the integration of an evidential deep learning method into a continual learning framework in order to perform simultaneously incremental object classification and OOD detection. Moreover, we analyze the ability of vacuity and dissonance to differentiate between in-distribution data belonging to old classes and OOD data. The proposed method, called CEDL, is evaluated on CIFAR-100 considering two settings consisting of 5 and 10 tasks, respectively. From the obtained results, we could appreciate that the proposed method, in addition to provide comparable results in object classification with respect to the baseline, largely outperforms OOD detection compared to several posthoc methods on three evaluation metrics: AUROC, AUPR and FPR95.

arXiv:2309.04081

UER: A Heuristic Bias Addressing Approach for Online Continual Learning

Online continual learning aims to continuously train neural networks from a continuous data stream with a single pass-through data. As the most effective approach, the rehearsal-based methods replay part of previous data. Commonly used predictors in existing methods tend to generate biased dot-product logits that prefer to the classes of current data, which is known as a bias issue and a phenomenon of forgetting. Many approaches have been proposed to overcome the forgetting problem by correcting the bias; however, they still need to be improved in online fashion. In this paper, we try to address the bias issue by a more straightforward and more efficient method. By decomposing the dot-product logits into an angle factor and a norm factor, we empirically find that the bias problem mainly occurs in the angle factor, which can be used to learn novel knowledge as cosine logits. On the contrary, the norm factor abandoned by existing methods helps remember historical knowledge. Based on this observation, we intuitively propose to leverage the norm factor to balance the new and old knowledge for addressing the bias. To this end, we develop a heuristic approach called unbias experience replay (UER). UER learns current samples only by the angle factor and further replays previous samples by both the norm and angle factors. Extensive experiments on three datasets show that UER achieves superior performance over various state-of-the-art methods. The code is in https://github.com/FelixHuiweiLin/UER.

arXiv:2309.04296공식 코드 · Papers with Code

Navigating Out-of-Distribution Electricity Load Forecasting during COVID-19: Benchmarking energy load forecasting models without and with continual learning

In traditional deep learning algorithms, one of the key assumptions is that the data distribution remains constant during both training and deployment. However, this assumption becomes problematic when faced with Out-of-Distribution periods, such as the COVID-19 lockdowns, where the data distribution significantly deviates from what the model has seen during training. This paper employs a two-fold strategy: utilizing continual learning techniques to update models with new data and harnessing human mobility data collected from privacy-preserving pedestrian counters located outside buildings. In contrast to online learning, which suffers from 'catastrophic forgetting' as newly acquired knowledge often erases prior information, continual learning offers a holistic approach by preserving past insights while integrating new data. This research applies FSNet, a powerful continual learning algorithm, to real-world data from 13 building complexes in Melbourne, Australia, a city which had the second longest total lockdown duration globally during the pandemic. Results underscore the crucial role of continual learning in accurate energy forecasting, particularly during Out-of-Distribution periods. Secondary data such as mobility and temperature provided ancillary support to the primary forecasting model. More importantly, while traditional methods struggled to adapt during lockdowns, models featuring at least online learning demonstrated resilience, with lockdown periods posing fewer challenges once armed with adaptive learning techniques. This study contributes valuable methodologies and insights to the ongoing effort to improve energy load forecasting during future Out-of-Distribution periods.

arXiv:2309.02870공식 코드 · Papers with Code

Rethinking Momentum Knowledge Distillation in Online Continual Learning

Online Continual Learning (OCL) addresses the problem of training neural networks on a continuous data stream where multiple classification tasks emerge in sequence. In contrast to offline Continual Learning, data can be seen only once in OCL, which is a very severe constraint. In this context, replay-based strategies have achieved impressive results and most state-of-the-art approaches heavily depend on them. While Knowledge Distillation (KD) has been extensively used in offline Continual Learning, it remains under-exploited in OCL, despite its high potential. In this paper, we analyze the challenges in applying KD to OCL and give empirical justifications. We introduce a direct yet effective methodology for applying Momentum Knowledge Distillation (MKD) to many flagship OCL methods and demonstrate its capabilities to enhance existing approaches. In addition to improving existing state-of-the-art accuracy by more than $10\%$ points on ImageNet100, we shed light on MKD internal mechanics and impacts during training in OCL. We argue that similar to replay, MKD should be considered a central component of OCL. The code is available at \url{https://github.com/Nicolas1203/mkd_ocl}.

arXiv:2309.03720공식 코드 · Papers with Code

A Natural Gas Consumption Forecasting System for Continual Learning Scenarios based on Hoeffding Trees with Change Point Detection Mechanism

Forecasting natural gas consumption, considering seasonality and trends, is crucial in planning its supply and consumption and optimizing the cost of obtaining it, mainly by industrial entities. However, in times of threats to its supply, it is also a critical element that guarantees the supply of this raw material to meet individual consumers' needs, ensuring society's energy security. This article introduces a novel multistep ahead forecasting of natural gas consumption with change point detection integration for model collection selection with continual learning capabilities using data stream processing. The performance of the forecasting models based on the proposed approach is evaluated in a complex real-world use case of natural gas consumption forecasting. We employed Hoeffding tree predictors as forecasting models and the Pruned Exact Linear Time (PELT) algorithm for the change point detection procedure. The change point detection integration enables selecting a different model collection for successive time frames. Thus, three model collection selection procedures (with and without an error feedback loop) are defined and evaluated for forecasting scenarios with various densities of detected change points. These models were compared with change point agnostic baseline approaches. Our experiments show that fewer change points result in a lower forecasting error regardless of the model collection selection procedure employed. Also, simpler model collection selection procedures omitting forecasting error feedback leads to more robust forecasting models suitable for continual learning tasks.

arXiv:2309.03487공식 코드 · Papers with Code

Privacy-preserving Continual Federated Clustering via Adaptive Resonance Theory

With the increasing importance of data privacy protection, various privacy-preserving machine learning methods have been proposed. In the clustering domain, various algorithms with a federated learning framework (i.e., federated clustering) have been actively studied and showed high clustering performance while preserving data privacy. However, most of the base clusterers (i.e., clustering algorithms) used in existing federated clustering algorithms need to specify the number of clusters in advance. These algorithms, therefore, are unable to deal with data whose distributions are unknown or continually changing. To tackle this problem, this paper proposes a privacy-preserving continual federated clustering algorithm. In the proposed algorithm, an adaptive resonance theory-based clustering algorithm capable of continual learning is used as a base clusterer. Therefore, the proposed algorithm inherits the ability of continual learning. Experimental results with synthetic and real-world datasets show that the proposed algorithm has superior clustering performance to state-of-the-art federated clustering algorithms while realizing data privacy protection and continual learning ability. The source code is available at \url{https://github.com/Masuyama-lab/FCAC}.

arXiv:2309.01811

Instant Continual Learning of Neural Radiance Fields

Neural radiance fields (NeRFs) have emerged as an effective method for novel-view synthesis and 3D scene reconstruction. However, conventional training methods require access to all training views during scene optimization. This assumption may be prohibitive in continual learning scenarios, where new data is acquired in a sequential manner and a continuous update of the NeRF is desired, as in automotive or remote sensing applications. When naively trained in such a continual setting, traditional scene representation frameworks suffer from catastrophic forgetting, where previously learned knowledge is corrupted after training on new data. Prior works in alleviating forgetting with NeRFs suffer from low reconstruction quality and high latency, making them impractical for real-world application. We propose a continual learning framework for training NeRFs that leverages replay-based methods combined with a hybrid explicit--implicit scene representation. Our method outperforms previous methods in reconstruction quality when trained in a continual setting, while having the additional benefit of being an order of magnitude faster.

arXiv:2309.00023

Continual Learning From a Stream of APIs

Continual learning (CL) aims to learn new tasks without forgetting previous tasks. However, existing CL methods require a large amount of raw data, which is often unavailable due to copyright considerations and privacy risks. Instead, stakeholders usually release pre-trained machine learning models as a service (MLaaS), which users can access via APIs. This paper considers two practical-yet-novel CL settings: data-efficient CL (DECL-APIs) and data-free CL (DFCL-APIs), which achieve CL from a stream of APIs with partial or no raw data. Performing CL under these two new settings faces several challenges: unavailable full raw data, unknown model parameters, heterogeneous models of arbitrary architecture and scale, and catastrophic forgetting of previous APIs. To overcome these issues, we propose a novel data-free cooperative continual distillation learning framework that distills knowledge from a stream of APIs into a CL model by generating pseudo data, just by querying APIs. Specifically, our framework includes two cooperative generators and one CL model, forming their training as an adversarial game. We first use the CL model and the current API as fixed discriminators to train generators via a derivative-free method. Generators adversarially generate hard and diverse synthetic data to maximize the response gap between the CL model and the API. Next, we train the CL model by minimizing the gap between the responses of the CL model and the black-box API on synthetic data, to transfer the API's knowledge to the CL model. Furthermore, we propose a new regularization term based on network similarity to prevent catastrophic forgetting of previous APIs.Our method performs comparably to classic CL with full raw data on the MNIST and SVHN in the DFCL-APIs setting. In the DECL-APIs setting, our method achieves 0.97x, 0.75x and 0.69x performance of classic CL on CIFAR10, CIFAR100, and MiniImageNet.

arXiv:2309.00462

New metrics for analyzing continual learners

Deep neural networks have shown remarkable performance when trained on independent and identically distributed data from a fixed set of classes. However, in real-world scenarios, it can be desirable to train models on a continuous stream of data where multiple classification tasks are presented sequentially. This scenario, known as Continual Learning (CL) poses challenges to standard learning algorithms which struggle to maintain knowledge of old tasks while learning new ones. This stability-plasticity dilemma remains central to CL and multiple metrics have been proposed to adequately measure stability and plasticity separately. However, none considers the increasing difficulty of the classification task, which inherently results in performance loss for any model. In that sense, we analyze some limitations of current metrics and identify the presence of setup-induced forgetting. Therefore, we propose new metrics that account for the task's increasing difficulty. Through experiments on benchmark datasets, we demonstrate that our proposed metrics can provide new insights into the stability-plasticity trade-off achieved by models in the continual learning environment.

arXiv:2308.16567공식 코드 · Papers with Code

ScrollNet: Dynamic Weight Importance for Continual Learning

The principle underlying most existing continual learning (CL) methods is to prioritize stability by penalizing changes in parameters crucial to old tasks, while allowing for plasticity in other parameters. The importance of weights for each task can be determined either explicitly through learning a task-specific mask during training (e.g., parameter isolation-based approaches) or implicitly by introducing a regularization term (e.g., regularization-based approaches). However, all these methods assume that the importance of weights for each task is unknown prior to data exposure. In this paper, we propose ScrollNet as a scrolling neural network for continual learning. ScrollNet can be seen as a dynamic network that assigns the ranking of weight importance for each task before data exposure, thus achieving a more favorable stability-plasticity tradeoff during sequential task learning by reassigning this ranking for different tasks. Additionally, we demonstrate that ScrollNet can be combined with various CL methods, including regularization-based and replay-based approaches. Experimental results on CIFAR100 and TinyImagenet datasets show the effectiveness of our proposed method. We release our code at https://github.com/FireFYF/ScrollNet.git.

arXiv:2308.15827공식 코드 · Papers with Code

Introducing Language Guidance in Prompt-based Continual Learning

Continual Learning aims to learn a single model on a sequence of tasks without having access to data from previous tasks. The biggest challenge in the domain still remains catastrophic forgetting: a loss in performance on seen classes of earlier tasks. Some existing methods rely on an expensive replay buffer to store a chunk of data from previous tasks. This, while promising, becomes expensive when the number of tasks becomes large or data can not be stored for privacy reasons. As an alternative, prompt-based methods have been proposed that store the task information in a learnable prompt pool. This prompt pool instructs a frozen image encoder on how to solve each task. While the model faces a disjoint set of classes in each task in this setting, we argue that these classes can be encoded to the same embedding space of a pre-trained language encoder. In this work, we propose Language Guidance for Prompt-based Continual Learning (LGCL) as a plug-in for prompt-based methods. LGCL is model agnostic and introduces language guidance at the task level in the prompt pool and at the class level on the output feature of the vision encoder. We show with extensive experimentation that LGCL consistently improves the performance of prompt-based continual learning methods to set a new state-of-the art. LGCL achieves these performance improvements without needing any additional learnable parameters.

arXiv:2308.14968공식 코드 · Papers with Code

Continual Learning for Generative Retrieval over Dynamic Corpora

Generative retrieval (GR) directly predicts the identifiers of relevant documents (i.e., docids) based on a parametric model. It has achieved solid performance on many ad-hoc retrieval tasks. So far, these tasks have assumed a static document collection. In many practical scenarios, however, document collections are dynamic, where new documents are continuously added to the corpus. The ability to incrementally index new documents while preserving the ability to answer queries with both previously and newly indexed relevant documents is vital to applying GR models. In this paper, we address this practical continual learning problem for GR. We put forward a novel Continual-LEarner for generatiVE Retrieval (CLEVER) model and make two major contributions to continual learning for GR: (i) To encode new documents into docids with low computational cost, we present Incremental Product Quantization, which updates a partial quantization codebook according to two adaptive thresholds; and (ii) To memorize new documents for querying without forgetting previous knowledge, we propose a memory-augmented learning mechanism, to form meaningful connections between old and new documents. Empirical results demonstrate the effectiveness and efficiency of the proposed model.