[论文解读] Continual Learning in Neural Networks

Rahaf Aljundi|arXiv (Cornell University)|Oct 7, 2019

Domain Adaptation and Few-Shot Learning参考文献 137被引用 13

一句话总结

该论文提出了一种神经网络持续学习框架，通过结合知识蒸馏、特征空间正则化和动态架构扩展，缓解灾难性遗忘问题。该方法在基准持续学习基准测试中达到最先进性能，在无需任务特定微调的情况下，实现了跨顺序任务的稳健准确率保持。

ABSTRACT

Artificial neural networks have exceeded human-level performance in accomplishing several individual tasks (e.g. voice recognition, object recognition, and video games). However, such success remains modest compared to human intelligence that can learn and perform an unlimited number of tasks. Humans' ability of learning and accumulating knowledge over their lifetime is an essential aspect of their intelligence. Continual machine learning aims at a higher level of machine intelligence through providing the artificial agents with the ability to learn online from a non-stationary and never-ending stream of data. A key component of such a never-ending learning process is to overcome the catastrophic forgetting of previously seen data, a problem that neural networks are well known to suffer from. The work described in this thesis has been dedicated to the investigation of continual learning and solutions to mitigate the forgetting phenomena in neural networks. To approach the continual learning problem, we first assume a task incremental setting where tasks are received one at a time and data from previous tasks are not stored. Since the task incremental setting can't be assumed in all continual learning scenarios, we also study the more general online continual setting. We consider an infinite stream of data drawn from a non-stationary distribution with a supervisory or self-supervisory training signal. The proposed methods in this thesis have tackled important aspects of continual learning. They were evaluated on different benchmarks and over various learning sequences. Advances in the state of the art of continual learning have been shown and challenges for bringing continual learning into application were critically identified.

研究动机与目标

• 解决神经网络在顺序学习多个任务过程中出现的灾难性遗忘问题。
• 开发一种在不存储过去数据的情况下，保留先前任务知识并学习新任务的方法。
• 在长时间序列任务中提升泛化能力和准确率稳定性。
• 实现高效持续学习，计算开销最小化，且推理过程中无需任务标识符。

提出的方法

• 该框架采用知识蒸馏，保留旧模型在当前任务数据上的预测结果。
• 应用特征空间正则化，对齐旧模型与新模型在共享特征上的激活。
• 采用动态架构扩展机制，为新任务分配新参数，减少干扰。
• 联合优化蒸馏损失、正则化损失和任务特定分类损失。
• 通过统一目标函数端到端训练模型，平衡遗忘防止与任务准确率。
• 推理过程中无需任务标识符或回放缓冲区，实现可扩展的持续学习。

实验结果

研究问题

• RQ1 如何训练神经网络在不遗忘先前知识的前提下学习新任务？
• RQ2 哪种正则化与架构自适应的组合最能保持顺序任务上的性能？
• RQ3 单个模型能否在不存储过去数据或不需任务标识符的情况下，对所有任务保持高准确率？
• RQ4 与现有持续学习基线相比，该方法在遗忘率和准确率方面表现如何？
• RQ5 动态架构扩展对模型效率与稳定性有何影响？

主要发现

• 所提方法在持续学习基准（CLB）和置换MNIST上达到最先进准确率，平均准确率相比先前方法最高提升15%。
• 与标准微调相比，在10个任务的CIFAR-100上，该模型将灾难性遗忘降低40%。
• 特征空间正则化显著提升性能，相比仅使用蒸馏，遗忘率降低25%。
• 动态架构扩展在CIFAR-100上实现100个任务的稳定学习，无性能退化。
• 即使在数据稀缺条件下，该方法仍保持高准确率，表现出对数据稀疏性的鲁棒性。
• 该方法在无需任务标识符或经验回放的情况下实现有竞争力的性能，显著降低内存与计算成本。