Deep Neural Networks (DNNs) have shown significant advantages in a wide
variety of domains. However, DNNs are becoming computationally intensive and
energy hungry at an exponential pace, while at the same time, there is a vast
demand for running sophisticated DNN-based services on resource constrained
embedded devices. In this paper, we target energy-efficient inference on
embedded DNN accelerators. To that end, we propose an automated framework to
compress DNNs in a hardware-aware manner by jointly employing pruning and
quantization. We explore, for the first time, per-layer fine- and
coarse-grained pruning, in the same DNN architecture, in addition to low
bit-width mixed-precision quantization for weights and activations.
Reinforcement Learning (RL) is used to explore the associated design space and
identify the pruning-quantization configuration so that the energy consumption
is minimized whilst the prediction accuracy loss is retained at acceptable
levels. Using our novel composite RL agent we are able to extract
energy-efficient solutions without requiring retraining and/or fine tuning. Our
extensive experimental evaluation over widely used DNNs and the CIFAR-10/100
and ImageNet datasets demonstrates that our framework achieves $39\%$ average
energy reduction for $1.7\%$ average accuracy loss and outperforms
significantly the state-of-the-art approaches.

通过联合采用修剪和量化，利用强化学习探索与低能耗相关的设计空间及其精度损失，我们提出了一种自动压缩深度神经网络的硬件感知框架，能够在嵌入式深度神经网络加速器上实现能耗最小化。实验结果表明，相比现有方法，我们的框架平均能耗降低 39%，平均精度损失为 1.7%。

硬件感知的 DNN 压缩：通过多样化剪枝和混合精度量化

Hardware-Aware DNN Compression via Diverse Pruning and Mixed-Precision  Quantization

To process sensor data in the Internet of Things(IoTs), embedded deep
learning for 1-dimensional data is an important technique. In the past, CNNs
were frequently used because they are simple to optimise for special embedded
hardware such as FPGAs. This work proposes a novel LSTM cell optimisation aimed
at energy-efficient inference on end devices. Using the traffic speed
prediction as a case study, a vanilla LSTM model with the optimised LSTM cell
achieves 17534 inferences per second while consuming only 3.8 $\mu$J per
inference on the FPGA \textit{XC7S15} from \textit{Spartan-7} family. It
achieves at least 5.4$\times$ faster throughput and 1.37$\times$ more energy
efficient than existing approaches.

为了处理物联网 (IoTs) 中的传感器数据，本研究提出了一种嵌入式深度学习的重要技术，即 LSTM 细胞的优化，旨在实现端设备上的高效能推断。通过以交通速度预测为案例研究，利用优化的 LSTM 细胞的简单 LSTM 模型在	extit {XC7S15} FPGA 芯片上实现了每秒 17534 次的推断，每次推断仅消耗 3.8 微焦耳的能量。与现有方法相比，其吞吐量至少提高了 5.4 倍，能量效率提高了 1.37 倍。