A number of recent efforts have attempted to design accelerators for popular machine learning algorithms, such as those involving convolutional and deep neural networks (CNNs and DNNs). These algorithms typically involve a large number of multiply-accumulate (dot-product) operations. A recent project, DaDianNao, adopts a near data processing approach, where a specialized neural functional unit performs all the digital arithmetic operations and receives input weights from adjacent eDRAM banks.This work explores an in-situ processing approach, where memristor crossbar arrays not only store input weights, but are also used to perform dot-product operations in an analog manner. While the use of crossbar memory as an analog dot-product engine is well known, no prior work has designed or characterized a full-fledged accelerator based on crossbars. In particular, our work makes the following contributions: (i) We design a pipelined architecture, with some crossbars dedicated for each neural network layer, and eDRAM buffers that aggregate data between pipeline stages. (ii) We define new data encoding techniques that are amenable to analog computations and that can reduce the high overheads of analog-to-digital conversion (ADC). (iii) We define the many supporting digital components required in an analog CNN accelerator and carry out a design space exploration to identify the best balance of memristor storage/compute, ADCs, and eDRAM storage on a chip. On a suite of CNN and DNN workloads, the proposed ISAAC architecture yields improvements of 14.8×, 5.5×, and 7.5× in throughput, energy, and computational density (respectively), relative to the state-of-the-art DaDianNao architecture.

ISAAC: a convolutional neural network accelerator with in-situ analog arithmetic in crossbars

With the breakthroughs in deep learning, the recent years have witnessed a booming of artificial intelligence (AI) applications and services, spanning from personal assistant to recommendation systems to video/audio surveillance. More recently, with the proliferation of mobile computing and Internet of Things (IoT), billions of mobile and IoT devices are connected to the Internet, generating zillions bytes of data at the network edge. Driving by this trend, there is an urgent need to push the AI frontiers to the network edge so as to fully unleash the potential of the edge big data. To meet this demand, edge computing, an emerging paradigm that pushes computing tasks and services from the network core to the network edge, has been widely recognized as a promising solution. The resulted new interdiscipline, edge AI or edge intelligence (EI), is beginning to receive a tremendous amount of interest. However, research on EI is still in its infancy stage, and a dedicated venue for exchanging the recent advances of EI is highly desired by both the computer system and AI communities. To this end, we conduct a comprehensive survey of the recent research efforts on EI. Specifically, we first review the background and motivation for AI running at the network edge. We then provide an overview of the overarching architectures, frameworks, and emerging key technologies for deep learning model toward training/inference at the network edge. Finally, we discuss future research opportunities on EI. We believe that this survey will elicit escalating attentions, stimulate fruitful discussions, and inspire further research ideas on EI.

/pdf/edge-intelligence-paving-the-last-mile-of-artificial-19sytmrpiq.pdf

Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing

The computation for today's intelligent personal assistants such as Apple Siri, Google Now, and Microsoft Cortana, is performed in the cloud. This cloud-only approach requires significant amounts of data to be sent to the cloud over the wireless network and puts significant computational pressure on the datacenter. However, as the computational resources in mobile devices become more powerful and energy efficient, questions arise as to whether this cloud-only processing is desirable moving forward, and what are the implications of pushing some or all of this compute to the mobile devices on the edge.In this paper, we examine the status quo approach of cloud-only processing and investigate computation partitioning strategies that effectively leverage both the cycles in the cloud and on the mobile device to achieve low latency, low energy consumption, and high datacenter throughput for this class of intelligent applications. Our study uses 8 intelligent applications spanning computer vision, speech, and natural language domains, all employing state-of-the-art Deep Neural Networks (DNNs) as the core machine learning technique. We find that given the characteristics of DNN algorithms, a fine-grained, layer-level computation partitioning strategy based on the data and computation variations of each layer within a DNN has significant latency and energy advantages over the status quo approach.Using this insight, we design Neurosurgeon, a lightweight scheduler to automatically partition DNN computation between mobile devices and datacenters at the granularity of neural network layers. Neurosurgeon does not require per-application profiling. It adapts to various DNN architectures, hardware platforms, wireless networks, and server load levels, intelligently partitioning computation for best latency or best mobile energy. We evaluate Neurosurgeon on a state-of-the-art mobile development platform and show that it improves end-to-end latency by 3.1X on average and up to 40.7X, reduces mobile energy consumption by 59.5% on average and up to 94.7%, and improves datacenter throughput by 1.5X on average and up to 6.7X.

https://dl.acm.org/doi/pdf/10.1145/3037697.3037698

Neurosurgeon: Collaborative Intelligence Between the Cloud and Mobile Edge

Deep learning is currently widely used in a variety of applications, including computer vision and natural language processing. End devices, such as smartphones and Internet-of-Things sensors, are generating data that need to be analyzed in real time using deep learning or used to train deep learning models. However, deep learning inference and training require substantial computation resources to run quickly. Edge computing, where a fine mesh of compute nodes are placed close to end devices, is a viable way to meet the high computation and low-latency requirements of deep learning on edge devices and also provides additional benefits in terms of privacy, bandwidth efficiency, and scalability. This paper aims to provide a comprehensive review of the current state of the art at the intersection of deep learning and edge computing. Specifically, it will provide an overview of applications where deep learning is used at the network edge, discuss various approaches for quickly executing deep learning inference across a combination of end devices, edge servers, and the cloud, and describe the methods for training deep learning models across multiple edge devices. It will also discuss open challenges in terms of systems performance, network technologies and management, benchmarks, and privacy. The reader will take away the following concepts from this paper: understanding scenarios where deep learning at the network edge can be useful, understanding common techniques for speeding up deep learning inference and performing distributed training on edge devices, and understanding recent trends and opportunities.

https://ieeexplore.ieee.org/ielaam/5/8789751/8763885-aam.pdf

Deep Learning With Edge Computing: A Review

Ubiquitous sensors and smart devices from factories and communities are generating massive amounts of data, and ever-increasing computing power is driving the core of computation and services from the cloud to the edge of the network. As an important enabler broadly changing people’s lives, from face recognition to ambitious smart factories and cities, developments of artificial intelligence (especially deep learning, DL) based applications and services are thriving. However, due to efficiency and latency issues, the current cloud computing service architecture hinders the vision of “providing artificial intelligence for every person and every organization at everywhere”. Thus, unleashing DL services using resources at the network edge near the data sources has emerged as a desirable solution. Therefore, edge intelligence , aiming to facilitate the deployment of DL services by edge computing, has received significant attention. In addition, DL, as the representative technique of artificial intelligence, can be integrated into edge computing frameworks to build intelligent edge for dynamic, adaptive edge maintenance and management. With regard to mutually beneficial edge intelligence and intelligent edge , this paper introduces and discusses: 1) the application scenarios of both; 2) the practical implementation methods and enabling technologies, namely DL training and inference in the customized edge computing framework; 3) challenges and future trends of more pervasive and fine-grained intelligence. We believe that by consolidating information scattered across the communication, networking, and DL areas, this survey can help readers to understand the connections between enabling technologies while promoting further discussions on the fusion of edge intelligence and intelligent edge , i.e., Edge DL.

Convergence of Edge Computing and Deep Learning: A Comprehensive Survey

As applications such as Apple Siri, Google Now, Microsoft Cortana, and Amazon Echo continue to gain traction, webservice companies are adopting large deep neural networks (DNN) for machine learning challenges such as image processing, speech recognition, natural language processing, among others. A number of open questions arise as to the design of a server platform specialized for DNN and how modern warehouse scale computers (WSCs) should be outfitted to provide DNN as a service for these applications. In this paper, we present DjiNN, an open infrastructure for DNN as a service in WSCs, and Tonic Suite, a suite of 7 end-to-end applications that span image, speech, and language processing. We use DjiNN to design a high throughput DNN system based on massive GPU server designs and provide insights as to the varying characteristics across applications. After studying the throughput, bandwidth, and power properties of DjiNN and Tonic Suite, we investigate several design points for future WSC architectures. We investigate the total cost of ownership implications of having a WSC with a disaggregated GPU pool versus a WSC composed of homogeneous integrated GPU servers. We improve DNN throughput by over 120× for all but one application (40× for Facial Recognition) on an NVIDIA K40 GPU. On a GPU server composed of 8 NVIDIA K40s, we achieve near-linear scaling (around 1000× throughput improvement) for 3 of the 7 applications. Through our analysis, we also find that GPU-enabled WSCs improve total cost of ownership over CPU-only designs by 4–20×, depending on the composition of the workload.

DjiNN and Tonic: DNN as a service and its implications for future warehouse scale computers

Industrial dialogue systems such as Apple Siri and Google Now rely on large scale diverse and robust training data to enable their sophisticated conversation capability. Crowdsourcing provides a scalable and inexpensive way of data collection but collecting high quality data efficiently requires thoughtful orchestration of the crowdsourcing jobs. Prior study of this topic have focused on tasks only in the academia settings with limited scope or only provide intrinsic dataset analysis, lacking indication on how it affects the trained model performance. In this paper, we present a study of crowdsourcing methods for a user intent classification task in our deployed dialogue system. Our task requires classification of 47 possible user intents and contains many intent pairs with subtle differences. We consider different crowdsourcing job types and job prompts and analyze quantitatively the quality of the collected data and the downstream model performance on a test set of real user queries from production logs. Our observation provides insights into designing efficient crowdsourcing jobs and provide recommendations for future dialogue system data collection process.

/pdf/data-collection-for-dialogue-system-a-startup-perspective-b0aisf5ojx.pdf

Data Collection for Dialogue System: A Startup Perspective

As user demand scales for intelligent personal assistants (IPAs) such as Apple’s Siri, Google’s Google Now, and Microsoft’s Cortana, we are approaching the computational limits of current datacenter (DC) architectures. It is an open question how future server architectures should evolve to enable this emerging class of applications, and the lack of an open-source IPA workload is an obstacle in addressing this question. In this article, we present the design of Sirius, an open end-to-end IPA Web-service application that accepts queries in the form of voice and images, and responds with natural language. We then use this workload to investigate the implications of four points in the design space of future accelerator-based server architectures spanning traditional CPUs, GPUs, manycore throughput co-processors, and FPGAs. To investigate future server designs for Sirius, we decompose Sirius into a suite of eight benchmarks (Sirius Suite) comprising the computationally intensive bottlenecks of Sirius. We port Sirius Suite to a spectrum of accelerator platforms and use the performance and power trade-offs across these platforms to perform a total cost of ownership (TCO) analysis of various server design points. In our study, we find that accelerators are critical for the future scalability of IPA services. Our results show that GPU- and FPGA-accelerated servers improve the query latency on average by 8.5× and 15×, respectively. For a given throughput, GPU- and FPGA-accelerated servers can reduce the TCO of DCs by 2.3× and 1.3×, respectively.

Designing Future Warehouse-Scale Computers for Sirius, an End-to-End Voice and Vision Personal Assistant

A data processing apparatus has a memory rename table for storing memory rename entries each identifying a mapping between a memory address of a location in memory and a mapped register of a plurality of registers. The mapped register is identified by a register number. In response to a store instruction, the store target memory address of the store instruction is mapped to a store destination register and so the data value is stored to the store destination register instead of memory. A memory rename entry is provided in the table to identify the mapping between the store target memory address and store destination target register. In response to a load instruction, if there is a hit in the memory rename table for the load target memory address then the loaded value can be read from the mapped register instead of memory.

Yiping Kang

Papers

Neurosurgeon: Collaborative Intelligence Between the Cloud and Mobile Edge

DjiNN and Tonic: DNN as a service and its implications for future warehouse scale computers

Data Collection for Dialogue System: A Startup Perspective

Designing Future Warehouse-Scale Computers for Sirius, an End-to-End Voice and Vision Personal Assistant

Data processing apparatus with memory rename table for mapping memory addresses to registers