… In this paper, we provide a comprehensive survey of recent advances in network acceleration,
compression, and accelerator design from both algorithm and hardware points of view. …

… neural network acceleration are focusing on providing an architecture for computing general
neural network… sparsity of the neural network through pruning the network properly. However…

… In this work, we propose a roofline-model-based method for convolutional neural network’s
FPGA acceleration. In this method we first optimize CNN’s computation and memory access. …

… Some companies have unveiled accelerators for deep learning inference … deep learning
training [10]. NVIDIA launches an open source project called NVIDIA Deep Learning Accelerator …

… Deep convolutional neural networks (CNNs) have gained … of parallelism in hardware
acceleration. We further put forward … solution that maximizes accelerator throughput under the …

Filter pruning has been widely applied to neural network compression and acceleration.
Existing methods usually utilize pre-defined pruning criteria, such as Lp-norm, to prune …

… Due to the recent advances in Deep Neural Network (DNN) … of networks are crucial
requirement of the DNN accelerator … CNN accelerator called SNACC (Scalable Neuro Accelerator …

… a CNN accelerator for … accelerator, called Eyeriss, that can support high throughput CNN
inference and optimizes for the energy efficiency of the entire system, including the accelerator …

… Abstract—Deep convolutional neural networks (CNNs) have achieved remarkable
performance at the cost of huge computation. As the CNN models become more complex and …

… layers between convolution layers and will be best suited for the proposed accelerator with
… We introduce the convolution loop acceleration strategy to reduce the on-/off-chip memory …