“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Interactive AI-powered services require low-latency evaluation of deep neural network (DNN) models—aka ""real-time AI"". The growing demand for computationally expensive, state-of-the-art DNNs, coupled with diminishing performance gains of general-purpose architectures, has fueled an explosion of specialized Neural Processing Units (NPUs). NPUs for interactive services should satisfy two requirements: (1) execution of DNN models with low latency, high throughput, and high efficiency, and (2) flexibility to accommodate evolving state-of-the-art models (e.g., RNNs, CNNs, MLPs) without costly silicon updates. This paper describes the NPU architecture for Project Brainwave, a production-scale system for real-time AI. The Brainwave NPU achieves more than an order of magnitude improvement in latency and throughput over state-of-the-art GPUs on large RNNs at a batch size of 1. The NPU attains this performance using a single-threaded SIMD ISA paired with a distributed microarchitecture capable of dispatching over 7M operations from a single instruction. The spatially distributed microarchitecture, scaled up to 96,000 multiply-accumulate units, is supported by hierarchical instruction decoders and schedulers coupled with thousands of independently addressable high-bandwidth on-chip memories, and can transparently exploit many levels of fine-grain SIMD parallelism. When targeting an FPGA, microarchitectural parameters such as native datapaths and numerical precision can be "synthesis specialized" to models at compile time, enabling atypically high FPGA performance competitive with hardened NPUs. When running on an Intel Stratix 10 280 FPGA, the Brainwave NPU achieves performance ranging from ten to over thirty-five teraflops, with no batching, on large, memory-intensive RNNs.

A configurable cloud-scale DNN processor for real-time AI

As a powerful tool, the vehicular network has been built to connect human communication and transportation around the world for many years to come. However, with the rapid growth of vehicles, the vehicular network becomes heterogeneous, dynamic, and large scaled, which makes it difficult to meet the strict requirements, such as ultralow latency, high reliability, high security, and massive connections of the next-generation (6G) network. Recently, machine learning (ML) has emerged as a powerful artificial intelligence (AI) technique to make both the vehicle and wireless communication highly efficient and adaptable. Naturally, employing ML into vehicular communication and network becomes a hot topic and is being widely studied in both academia and industry, paving the way for the future intelligentization in 6G vehicular networks. In this article, we provide a survey on various ML techniques applied to communication, networking, and security parts in vehicular networks and envision the ways of enabling AI toward a future 6G vehicular network, including the evolution of intelligent radio (IR), network intelligentization, and self-learning with proactive exploration.

Future Intelligent and Secure Vehicular Network Toward 6G: Machine-Learning Approaches

Binary Neural Networks: A Survey

Pruning and quantization for deep neural network acceleration: A survey

Convolutional neural networks (CNN) are the current stateof-the-art for many computer vision tasks. CNNs outperform older methods in accuracy, but require vast amounts of computation and memory. As a result, existing CNN applications are typically run on clusters of CPUs or GPUs. Studies into the FPGA acceleration of CNN workloads has achieved reductions in power and energy consumption. However, large GPUs outperform modern FPGAs in throughput, and the existence of compatible deep learning frameworks give GPUs a significant advantage in programmability. Recent research in machine learning demonstrates the potential of very low precision CNNs -- i.e., CNNs with binarized weights and activations. Such binarized neural networks (BNNs) appear well suited for FPGA implementation, as their dominant computations are bitwise logic operations and their memory requirements are reduced. A combination of low-precision networks and high-level design methodology may help address the performance and productivity gap between FPGAs and GPUs. In this paper, we present the design of a BNN accelerator that is synthesized from C++ to FPGA-targeted Verilog. The accelerator outperforms existing FPGA-based CNN accelerators in GOPS as well as energy and resource efficiency.

https://dl.acm.org/doi/pdf/10.1145/3020078.3021741

Accelerating Binarized Convolutional Neural Networks with Software-Programmable FPGAs

Multi-hop wireless broadcast is an important component of vehicular networks. Many services rely on the performance of broadcast communication to disseminate data packets. In an urban vehicular network, an efficient way of broadcasting data packets is choosing some vehicles as relay nodes. The base station only needs to multicast data packets to these relay vehicles and the packets would be spread to the whole network through D2D communication among vehicles. In this situation, the strategy of relay selection becomes a key factor of the broadcast efficiency. In this paper, we provide an unsupervised-learning-based method developed from k-means algorithm to help select relay nodes. The base station can learn the distribution of devices itself and choose the relay devices automatically. We use Manhattan map as our map model for the simulation and the result shows great efficiency over a random- selecting strategy.

An Unsupervised-Learning-Based Method for Multi-Hop Wireless Broadcast Relay Selection in Urban Vehicular Networks

3D teeth reconstruction from X-ray is important for dental diagnosis and many clinical operations. However, no existing work has explored the reconstruction of teeth for a whole cavity from a single panoramic radiograph. Different from single object reconstruction from photos, this task has the unique challenge of constructing multiple objects at high resolutions. To conquer this task, we develop a novel ConvNet X2Teeth that decomposes the task into teeth localization and single-shape estimation. We also introduce a patch-based training strategy, such that X2Teeth can be end-to-end trained for optimal performance. Extensive experiments show that our method can successfully estimate the 3D structure of the cavity and reflect the details for each tooth. Moreover, X2Teeth achieves a reconstruction IoU of 0.681, which significantly outperforms the encoder-decoder method by \(1.71{\times }\) and the retrieval-based method by \(1.52{\times }\). Our method can also be promising for other multi-anatomy 3D reconstruction tasks.

X2Teeth: 3D Teeth Reconstruction from a Single Panoramic Radiograph

Label propagation is a popular technique for anatomical segmentation. In this work, we propose a novel deep framework for label propagation based on non-local label fusion. Our framework, named CompareNet, incorporates subnets for both extracting discriminating features, and learning the similarity measure, which lead to accurate segmentation. We also introduce the voxel-wise classification as an unary potential to the label fusion function, for alleviating the search failure issue of the existing non-local fusion strategies. Moreover, CompareNet is end-to-end trainable, and all the parameters are learnt together for the optimal performance. By evaluating CompareNet on two public datasets IBSRv2 and MICCAI 2012 for brain segmentation, we show it outperforms state-of-the-art methods in accuracy, while being robust to pathologies.

Weinan Song

Papers

Accelerating Binarized Convolutional Neural Networks with Software-Programmable FPGAs

An Unsupervised-Learning-Based Method for Multi-Hop Wireless Broadcast Relay Selection in Urban Vehicular Networks

X2Teeth: 3D Teeth Reconstruction from a Single Panoramic Radiograph

CompareNet: Anatomical Segmentation Network with Deep Non-local Label Fusion

Superior acetone sensor based on hetero-interface of SnSe2/SnO2 quasi core shell nanoparticles for previewing diabetes.