MXene-based optoelectronic synaptic transistors utilize attentional mechanisms to achieve hierarchical responses

Faced with a huge amount of information, the brain relies on attention mechanisms to highly select information for efficient processing. The degree of processing capacity required for information in different scenarios also varies, and the brain relies on attention to significantly enhance or weaken its ability to process information. However, many studies on attention only focus on the behavioral differences between attention and no attention, without further research on the degree of regulation of attention. The application scenarios of artificial vision systems are so complex that we need to regulate the degree of attention to cope with different information processing intensities. In this work, we demonstrated an optoelectronic synaptic transistor based on the mixture of PDVT-10 and MXene-TiO2, and for the first time imitated the mechanism of attention regulation signals at the biological level. Based on the attention mechanism, for slow-moving objects, we enhance the ability of data processing to prevent data loss caused by undersampling, and for fast-moving objects, we weaken the ability of data processing to prevent data redundancy caused by oversampling. In addition, our device array determines the location of moving targets and sends them to the YOLO network for dynamic target detection. Our research results show that the device realizes a hierarchical response to adapt to objects with different motion speeds, which expands the application scenarios of optoelectronic synapses. Faced with a huge amount of information, the brain relies on attention mechanisms to highly select information for efficient processing.