Event-based sensors are drawing increasing atten?tion due to their high temporal resolution, low power con?sumption, and low bandwidth. To efficiently extract semantically meaningful information from sparse data streams produced by such sensors, we present a 4.5TOP/s/W digital accelerator capable of performing 4-bits-quantized event-based convolutional neural networks (eCNN). Compared to standard convolutional engines, our accelerator performs a number of operations proportional to the number of events contained into the input data stream, ultimately achieving a high energy-to-information processing proportionality. On the IBM-DVS-Gesture dataset, we report 80uJ/inf to 261uJ/inf, respectively, when the input activity is 1.2% and 4.9%. Our accelerator consumes 0.221pJ/SOP, to the best of our knowledge it is the lowest energy/OP reported on a digital neuromorphic engine.
翻译:事件感应器由于高时间分辨率、低功率共振率和低带宽度而正在上升 。 为了高效地从这种感应器产生的稀散数据流中提取具有地震意义的信息,我们提供了一种4.5TOP/s/W数字加速器,能够进行4位数的量化事件立体神经网络(eCNN)。 与标准革命引擎相比,我们的加速器进行了一系列与输入数据流中包含的事件数量成比例的操作,最终达到高能量对信息处理比例。在IBM-DVS-Gesture数据集上,我们分别向261/J/inf报告80uJ/inf,而输入活动为1.2%和4.9%。我们的感应器消耗0.221pJ/SOP,据我们所知,它是数字神经形态引擎上报告的最低能量/操作量。