Interactive Continual Learning: Fast and Slow Thinking

TL;DR

This paper presents a novel Interactive Continual Learning framework, enabled by collaborative interactions among models of various sizes, and introduces the von Mises-Fisher Outlier Detection and Interaction strategy to identify hard examples, thus enhancing collaboration between System1 and System2 for complex reasoning realization.

Abstract

Advanced life forms, sustained by the synergistic interaction of neural cognitive mechanisms, continually acquire and transfer knowledge throughout their lifespan. In contrast, contemporary machine learning paradigms exhibit limitations in emulating the facets of continual learning (CL). Nonetheless, the emergence of large language models (LLMs) presents promising avenues for realizing CL via interactions with these models. Drawing on Complementary Learning System theory, this paper presents a novel Interactive Continual Learning (ICL) framework, enabled by collaborative interactions among models of various sizes. Specifically, we assign the ViT model as System1 and multimodal LLM as System2. To enable the memory module to deduce tasks from class information and enhance Set2Set retrieval, we propose the Class-Knowledge-Task Multi-Head Attention (CKT-MHA). Additionally, to improve memory retrieval in System1 through enhanced geometric representation, we introduce the CL-vMF mechanism, based on the von Mises-Fisher (vMF) distribution. Meanwhile, we introduce the von Mises-Fisher Outlier Detection and Interaction (vMF-ODI) strategy to identify hard examples, thus enhancing collaboration between System1 and System2 for complex reasoning realization. Comprehensive evaluation of our proposed ICL demonstrates significant resistance to forgetting and superior performance relative to existing methods. Code is available at github.com/ICL.

Figures (2)

• Figure 1: Comprehensive Training and Testing Illustration. In the training phase: we propose the CKT-MHA unified storage module for System1. And then use memory selection and updates through our CL-VMF mechanism with the EM strategy to optimize CL for the small model ViT. In the inference phase: 1) The process begins by assessing sample complexity using proposed vMF-ODI in System 1. 2) The System1 then swiftly generates inferential predictions. 3) If test samples surpass a complexity threshold, we activate collaborative inference. Specifically, the predictive results from System1 is used as background knowledge to narrow the scope of inference. 4) Subsequently, complex reasoning through the multimodal LLM is applied to achieve the final prediction.
• Figure 2: Further analysis of the proposed ICL. (a) and (b) are forgetting curves of different methods on the ImageNet-R in the Class IL and Task IL scenario respectively. (c) and (d) are the impact of regularization parameters $\lambda$ and $\delta$ on CIFAR100 and ImageNet-R respectively. (e) The impact of concentration $\kappa$, is evaluated at values of 0.5, 1, 1.5, and 2, respectively. (f) The impact of the number of category choices $K$ in the prompt, is evaluated at values of 2, 3, 4, and 5 respectively. (g) The impact of training memory without EM strategy. (h) The impact of training memory with EM strategy.