Modeling learnable electrical synapse for high precision spatio-temporal recognition

Wu, Zhenzhi; Zhang, Zhihong; Gao, Huanhuan; Jun, Qin; Zhao, Rongzhen; Zhao, Guiping; Li, Guoqi

doi:10.1016/j.neunet.2022.02.006

Cited by 4 publications

(4 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The classified output is obtained through the fully connected (FC) layer. In version 2, we refine the network by substituting LIAF with the electrical coupling LIAF-RMP neural model (Wu et al, 2022b ). Although the LIAF-RMP model achieves better accuracy, it incurs higher computational cost and a larger parameter size (the electrical synapses lead to a network weight size increase from 12 × 10 6 to 24 × 10 6 ).…”

Section: Methodsmentioning

confidence: 99%

“…The definition of LIF, Leaky Integrate and Analog Fire (LIAF), and Residual Membrane Potential (RMP) applied in this study has been previously illustrated in Han et al ( 2020 ), Wu et al ( 2021 ), and Wu et al ( 2022b ). For an easy understanding of the proposed framework, we reintroduced the definitions as follows.…”

Section: Generalized Spatiotemporal Processing Via Neural Dynamicsmentioning

confidence: 99%

“…These perceptions require high processing precision and need to be performed in real time with low computational power requirements. It is worth noting that brain-inspired computing features rich neural dynamics for efficiently processing temporal signals (Carlos et al, 2019 ) and deep learning technologies provide the capability for high-precision spatial information processing, which leads to the widespread development of the deep spiking neural networks (DSNNs) be deployed for spatiotemporal processing (Gu et al, 2019 ; Wu et al, 2021 , 2022b ). However, most of the current DSNN research focuses on dynamic vision sensor (DVS) processing or image recognition tasks.…”

Section: Introductionmentioning

confidence: 99%

“…. neural networks (DSNNs) be deployed for spatiotemporal processing (Gu et al, 2019;Wu et al, 2021Wu et al, , 2022b. However, most of the current DSNN research focuses on dynamic vision sensor (DVS) processing or image recognition tasks.…”

mentioning

confidence: 99%

See 3 more Smart Citations

BIDL: a brain-inspired deep learning framework for spatiotemporal processing

Wu¹,

Shen²,

Zhang³

et al. 2023

Front. Neurosci.

Self Cite

View full text Add to dashboard Cite

Brain-inspired deep spiking neural network (DSNN) which emulates the function of the biological brain provides an effective approach for event-stream spatiotemporal perception (STP), especially for dynamic vision sensor (DVS) signals. However, there is a lack of generalized learning frameworks that can handle various spatiotemporal modalities beyond event-stream, such as video clips and 3D imaging data. To provide a unified design flow for generalized spatiotemporal processing (STP) and to investigate the capability of lightweight STP processing via brain-inspired neural dynamics, this study introduces a training platform called brain-inspired deep learning (BIDL). This framework constructs deep neural networks, which leverage neural dynamics for processing temporal information and ensures high-accuracy spatial processing via artificial neural network layers. We conducted experiments involving various types of data, including video information processing, DVS information processing, 3D medical imaging classification, and natural language processing. These experiments demonstrate the efficiency of the proposed method. Moreover, as a research framework for researchers in the fields of neuroscience and machine learning, BIDL facilitates the exploration of different neural models and enables global-local co-learning. For easily fitting to neuromorphic chips and GPUs, the framework incorporates several optimizations, including iteration representation, state-aware computational graph, and built-in neural functions. This study presents a user-friendly and efficient DSNN builder for lightweight STP applications and has the potential to drive future advancements in bio-inspired research.

show abstract

Section: Methodsmentioning

confidence: 99%