Optimizing deep video representation to match brain activity

利用脑波进行动态难度调整以优化参与度

Dynamic Difficulty Adjustment With Brain Waves as a Tool for Optimizing Engagement

本研究利用脑电图脑波监测实现虚拟现实游戏中的动态难度调整，通过任务参与指数实时调整挑战级别，显著提升参与度19.79%，并具有自适应学习等应用潜力。

通过人类脑电表征对齐实现更类人脑的视觉

Achieving More Human Brain-Like Vision via Human EEG Representational Alignment

针对现有目标识别模型模拟人脑视觉处理不足、过往研究依赖非人类侵入式记录的问题，提出基于非侵入式EEG的视觉模型ReAlnet及多层编码框架，提升了模型与人脑表征的相似度，推动类脑AI发展。

BHN: A Brain-like Heterogeneous Network

The human brain works in an unsupervised way, and more than one brain region is essential for lighting up intelligence. Inspired by this, we propose a brain-like heterogeneous network (BHN), which can cooperatively learn a lot of distributed representations and one global attention representation. By optimizing distributed, self-supervised, and gradient-isolated objective functions in a minimax fashion, our model improves its representations, which are generated from patches of pictures or frames of videos in experiments.

利用反馈-前馈对齐实现类脑灵活视觉推理

Brain-like Flexible Visual Inference by Harnessing Feedback-Feedforward Alignment

本文提出反馈-前馈对齐算法FFA，通过前馈与反馈通路互为信用分配图实现协同优化，在MNIST和CIFAR10上赋予反馈连接去噪、去遮挡、幻觉与想象等灵活视觉推理能力，并缓解反向传播的权重传输问题，提升生物合理性。

Wireless Sensor Network based Future of Telecom Applications

A system and method for enabling human beings to communicate by way of their monitored brain activity. The brain activity of an individual is monitored and transmitted to a remote location (e.g. by satellite). At the remote location, the monitored brain activity is compared with pre-recorded normalized brain activity curves, waveforms, or patterns to determine if a match or substantial match is found. If such a match is found, then the computer at the remote location determines that the individual was attempting to communicate the word, phrase, or thought corresponding to the matched stored no…

神经形态能量感知学习用于自适应深部脑刺激

Neuromorphic Energy-Aware Learning for Adaptive Deep Brain Stimulation

提出能量感知学习，将执行器能耗纳入强化学习奖励，在帕金森病深部脑刺激中抑制病理振荡45.2%并减少刺激电荷80.0%，通过稀疏知识蒸馏部署到神经形态处理器，推理能耗降低28.1倍。