科研成果 by Type: Conference Paper

Spiking Neural Networks (SNNs) have been attached great importance due to their biological plausibility and high energy-efficiency on neuromorphic chips. As these chips are usually resource-constrained, the compression of SNNs is thus crucial along the road of practical use of SNNs. Most existing methods directly apply pruning approaches in artificial neural networks (ANNs) to SNNs, which ignore the difference between ANNs and SNNs, thus limiting the performance of the pruned SNNs. Besides, these methods are only suitable for shallow SNNs. In this paper, inspired by synaptogenesis and synapse elimination in the neural system, we propose gradient rewiring (Grad R), a joint learning algorithm of connectivity and weight for SNNs, that enables us to seamlessly optimize network structure without retraining. Our key innovation is to redefine the gradient to a new synaptic parameter, allowing better exploration of network structures by taking full advantage of the competition between pruning and regrowth of connections. The experimental results show that the proposed method achieves minimal loss of SNNs' performance on MNIST and CIFAR-10 datasets so far. Moreover, it reaches a ~3.5% accuracy loss under unprecedented 0.73% connectivity, which reveals remarkable structure refining capability in SNNs. Our work suggests that there exists extremely high redundancy in deep SNNs. Our codes are available at https://github.com/Yanqi-Chen/Gradient-Rewiring.

Recently, a novel retina-inspired camera, namely spike camera, has shown great potential for recording high-speed dynamic scenes. Unlike the conventional digital cameras that compact the visual information within the exposure interval into a single snapshot, the spike camera continuously outputs binary spike streams to record the dynamic scenes, yielding a very high temporal resolution. Most of the existing reconstruction methods for spike camera focus on reconstructing images with the same resolution as spike camera. However, as a trade-off of high temporal resolution, the spatial resolution of spike camera is limited, resulting in inferior details of the reconstruction. To address this issue, we develop a spike camera super-resolution framework, aiming to super resolve high-resolution intensity images from the low-resolution binary spike streams. Due to the relative motion between the camera and the objects to capture, the spikes fired by the same sensor pixel no longer describes the same points in the external scene. In this paper, we properly exploit the relative motion and derive the relationship between light intensity and each spike, so as to recover the external scene with both high temporal and high spatial resolution. Experimental results demonstrate that the proposed method can reconstruct pleasant high-resolution images from low-resolution spike streams.

In passive millimeter wave radiation images, reflection will affect the analysis of the target and scene. Based on the theoretical model of reflection, this paper analyzes the change of brightness temperature(TB) in the target region and the reflected region under specific scenes. A target detection method based on the change of TB is proposed to distinguish the target and its reflection region and obtain the target outline. The effectiveness of the method is verified by simulation and imaging experiments.

Koinon assemblies during the Principate were occasions when member communities dispatch delegations to gather at specifically designated cities, where they elect koinon officers and deliberate a range of affairs –– chiefly among which were festivities and sacrifices that honored the Roman emperors and the local cultic and civic traditions, but also revenues and expenditures, administrative tasks delegated by the imperial government, among others (Deininger 1965: 137-147; Edelmann-Singer 2015: 193-248, 309-310). Much has been discussed regarding the institutional aspects of the koinon assemblies; what could benefit from more discussion is the act of traveling to koinon assemblies. This paper assembles a small number of literary and epigraphic references that provide circumstantial references to koinon assembly-related travel anecdotes. Of particular importance among these are Strabo's description of the gathering of delegates from Lycian cities to the koinon meeting each year (Strab. 13.3.3), Aelius Aristides' account of the city of Smyrna's manipulative nomination of him as a candidate for the high priest of Asia (Ael. Arist. Hieroi Logoi 4.99-104), and the inscription honoring Quintus Popilius Python's gift to attendees of the koinon assembly while serving as the high priest of the Macedonian koinon (EKM 117). By assembling these and other evidence, this paper wishes to suggest that koinon assemblies were compulsory events that each member community would have to participate in, often at their own expense. Wealthy koinon-officeholders may opt to offset the burdens that communities (or their designated representatives) would have to shoulder in dispatching delegation, and such benefaction may be viewed from the perspective of a soft mobilization of the provincial elites in order to facilitate the orderly execution of business in the interest of the public weal.