[论文解读] Advancing Perception in Artificial Intelligence through Principles of Cognitive Science

Palaash Agrawal, Cheston Tan+1 more|arXiv (Cornell University)|Oct 13, 2023

Neural Networks and ApplicationsComputer Science被引用 3

一句话总结

本文提出将认知科学原理——源自神经科学、心理学和语言学——整合到人工智能中，以推进感知系统，旨在提升鲁棒性、泛化能力和效率。通过将认知理论映射到人工智能技术，作者识别出现有AI感知系统中的关键差距，并倡导发展具身化、可解释性和社会智能的系统，以推动人工通用智能的发展。

ABSTRACT

Although artificial intelligence (AI) has achieved many feats at a rapid pace, there still exist open problems and fundamental shortcomings related to performance and resource efficiency. Since AI researchers benchmark a significant proportion of performance standards through human intelligence, cognitive sciences-inspired AI is a promising domain of research. Studying cognitive science can provide a fresh perspective to building fundamental blocks in AI research, which can lead to improved performance and efficiency. In this review paper, we focus on the cognitive functions of perception, which is the process of taking signals from one's surroundings as input, and processing them to understand the environment. Particularly, we study and compare its various processes through the lens of both cognitive sciences and AI. Through this study, we review all current major theories from various sub-disciplines of cognitive science (specifically neuroscience, psychology and linguistics), and draw parallels with theories and techniques from current practices in AI. We, hence, present a detailed collection of methods in AI for researchers to build AI systems inspired by cognitive science. Further, through the process of reviewing the state of cognitive-inspired AI, we point out many gaps in the current state of AI (with respect to the performance of the human brain), and hence present potential directions for researchers to develop better perception systems in AI.

研究动机与目标

• 解决当前人工智能感知系统的基本局限，例如脆弱性、数据效率低下和泛化能力差。
• 通过借鉴认知科学理论，弥合人类感知与人工感知之间的差距。
• 识别阻碍人工智能感知向人工通用智能发展的开放性问题。
• 通过提出关键研究方向——泛化能力、具身认知、可解释性和社会智能——为未来研究提供指导。
• 推动开发不仅能力更强，而且更符合人类认知过程的人工智能系统。

提出的方法

• 系统性回顾认知科学各子领域——神经科学、心理学和语言学——在感知方面的核心理论。
• 将认知科学原理映射到计算机视觉和自然语言处理中的现有人工智能技术。
• 识别人类感知机制与深度学习架构之间的结构与功能相似性。
• 提出认知启发的设计应避免任务特定的工程设计，而应强调可泛化的、生物学上合理的机制。
• 建议使用心理物理学数据和认知图式来正则化人工智能模型，以提升感知能力和迁移学习效果。
• 倡导整合元学习和可解释人工智能框架，以增强对模型行为的可解释性和控制力。

Figure 1: Cognitive-AI is the intersection of cognitive sciences and artificial intelligence. Cognitive sciences itself is a multidisciplinary field, and the intersection of all these fields with AI (the green region) is of great importance to understand intelligence. We study the state of cognitive

实验结果

研究问题

• RQ1 认知科学理论如何提升人工智能感知系统的鲁棒性和效率？
• RQ2 当前人工智能感知中的关键差距是什么，导致其在任务和领域间无法实现泛化？
• RQ3 具身认知和环境交互在何种方式下可超越静态数据输入，增强人工智能感知能力？
• RQ4 如何系统性地将可解释性和可解释性整合到感知模型中，利用认知图式？
• RQ5 社会智能以及伦理推理在塑造与人类认知一致的未来人工智能感知系统中扮演什么角色？

主要发现

• 当前的人工智能感知系统仍显脆弱，尤其在噪声、对抗性或分布外输入下表现不佳，表明其鲁棒性不足。
• 少样本学习和长尾识别的局限性凸显了深度学习模型在数据需求上的低效性，与人类感知相比存在明显差距。
• 领域差距和负迁移现象阻碍了迁移学习的进展，表明需要基于认知的正则化方法以提升知识迁移效果。
• 具身人工智能仍局限于视觉导航等狭窄任务，凸显了需要认知原则来引导更丰富的环境交互。
• 深度学习的黑箱特性阻碍了信任与控制，迫切需要可解释人工智能和元学习框架以提升可解释性。
• 社会智能，包括伦理推理和文化细微差别，对于确保人工智能系统在现实环境中安全、负责任地运行至关重要。

Figure 2: Processes of perception in the brain. Perception is the collective outcome of five major processes – sensory stimulation, modularity and multisensory integration, bottom-up processing, top-down processing and interpretation. For each of these processes, we study concepts from neuroscience,