Artificial Intelligence Processing Research Articles

A Novel Instruction Driven 1-D CNN Processor for ECG Classification.

Electrocardiography (ECG) has emerged as a ubiquitous diagnostic tool for the identification and characterization of diverse cardiovascular pathologies. Wearable health monitoring devices, equipped with on-device biomedical artificial intelligence (AI) processors, have revolutionized the acquisition, analysis, and interpretation of ECG data. However, these systems necessitate AI processors that exhibit flexible configuration, facilitate portability, and demonstrate optimal performance in terms of power consumption and latency for the realization of various functionalities. To address these challenges, this study proposes an instruction-driven convolutional neural network (CNN) processor. This processor incorporates three key features: (1) An instruction-driven CNN processor to support versatile ECG-based application. (2) A Processing element (PE) array design that simultaneously considers parallelism and data reuse. (3) An activation unit based on the CORDIC algorithm, supporting both Tanh and Sigmoid computations. The design has been implemented using 110 nm CMOS process technology, occupying a die area of 1.35 mm2 with 12.94 µW power consumption. It has been demonstrated with two typical ECG AI applications, including two-class (i.e., normal/abnormal) classification and five-class classification. The proposed 1-D CNN algorithm performs with a 97.95% accuracy for the two-class classification and 97.9% for the five-class classification, respectively.

Smart heart disease prediction and amalgamation tracking system

In this modern world, cardiovascular disease stands as the leading cause of global mortality. To combat this alarming trend and prevent the devastating loss of lives, an innovative solution that focus on reliability, accuracy, scalability, and cost-effectiveness. This work proposes a system that utilizes an artificial intelligence processor (LSAI48266X) and an IoT device to transfer data from sensors such as the Mach30100 and DS18B20. The system aims to track, visualize, and forecast heart disease. Random Forest is a machine learning algorithm that predicts cardiac illness based on numerous parameters such as SpO2, heartbeat, temperature, and blood pressure. Web application is developed using PHP that displays hospital details and integrated with a Telegram chatbot for communication during emergency conditions. Compared to earlier methods, our proposed system distinguishes itself with an impressive accuracy rate of 95.6% and automation system to prevent loss of human life by incorporating the Random Forest Algorithm and tracking system.

Simulating an Integrated Photonic Image Classifier for Diffractive Neural Networks.

The slowdown of Moore's law and the existence of the "von Neumann bottleneck" has led to electronic-based computing systems under von Neumann's architecture being unable to meet the fast-growing demand for artificial intelligence computing. However, all-optical diffractive neural networks provide a possible solution to this challenge. They can outperform conventional silicon-based electronic neural networks due to the significantly higher speed of the propagation of optical signals (≈108 m.s-1) compared to electrical signals (≈105 m.s-1), their parallelism in nature, and their low power consumption. The integrated diffractive deep neural network (ID2NN) uses an on-chip fully passive photonic approach to achieve the functionality of neural networks (matrix-vector operations) and can be fabricated via the CMOS process, which is technologically more amenable to implementing an artificial intelligence processor. In this paper, we present a detailed design framework for the integrated diffractive deep neural network and corresponding silicon-on-insulator integration implementation through Python-based simulations. The performance of our proposed ID2NN was evaluated by solving image classification problems using the MNIST dataset.

A Multimodel Edge Computing Offloading Framework for Deep-Learning Application Based on Bayesian Optimization

With the rapid development of Internet of Things (IoT), data generated by IoT devices are also increasing exponentially. The edge computing has alleviated the problems of limited network and transmission delay when processing tasks of IoT devices in traditional cloud computing. And with the popularity of deep learning, more and more terminal devices are embedded with AI (Artificial Intelligence) processors for higher processing capability at the edge. However, the problems of Deep Learning task offloading in heterogeneous edge computing environment have not been fully investigated. In this paper, a multi-model edge computing offloading framework is proposed, using NVIDIA Jetson edge devices (Jetson TX2, Jetson Xavier NX, and Jetson Nano) and GeForce RTX GPU servers (RTX3080 and RTX2080) to simulate the edge computing environment, and make binary computational offloading decisions for face detection tasks. We also introduce a Bayesian Optimization algorithm, namely MTPE (Modified Tree-structured Parzen Estimator), to reduce the total cost of edge computation within a time slot including response time and energy consumption, and ensure the accuracy requirements of face detection. In addition, we employ the Lyapunov model to obtain the harvesting energy between time slots to keep the energy queue stable. Experiments reveal that MTPE algorithm can achieve the globally optimal solution in fewer iterations. The total cost of multi-model edge computing framework is reduced by an average of 17.94% compared to a single-model framework. In contrast to the Double Deep Q-Network (DDQN), our proposed algorithm can decrease the computational consumption by 23.01% for obtaining the offloading decision.

Retracted: Obstructive Sleep Apnea Syndrome Treated Using a Positive Pressure Ventilator Based on Artificial Intelligence Processor.

[This retracts the article DOI: 10.1155/2021/5683433.].

Benchmarking Deep Learning Models on Myriad and Snapdragon Processors for Space Applications

Future space missions can benefit from processing imagery on board to detect science events, create insights, and respond autonomously. One of the challenges to this mission concept is that traditional space flight computing has limited capabilities because it is derived from much older computing to ensure reliable performance in the extreme environments of space: particularly radiation. Modern commercial-off-the-shelf processors, such as the Movidius Myriad X and the Qualcomm Snapdragon, provide significant improvements in small size, weight, and power packaging; and they offer direct hardware acceleration for deep neural networks, although these processors are not radiation hardened. We deploy neural network models on these processors hosted by Hewlett Packard Enterprise’s Spaceborne Computer-2 on board the International Space Station (ISS). We find that the Myriad and Snapdragon digital signal processors (DSP)/artificial intelligence processors (AIP) provide speed improvement over the Snapdragon CPU in all cases except single-pixel networks (typically greater than 10 times for DSP/AIP). In addition, the discrepancy introduced through quantization and porting of our Jet Propulsion Laboratory models was usually quite low (less than 5%). Models were run multiple times, and memory checkers were deployed to test for radiation effects. To date, we have found no difference in output between ground and ISS runs, and no memory checker errors.

From Algorithm to Module: Adaptive and Energy-Efficient Quantization Method for Edge Artificial Intelligence in IoT Society

Next-generation industrial edge artificial intelligence (AI) applications will undoubtedly emerge on energy-efficient, highly-integrated platforms incorporating various sensors, processors, and functional modules. Effectively bridging sensor data captured from a microprocessor or central processing unit with dedicated processing modules is still being studied in academia and industry. Therefore, embedding the data quantization module onto edge AI chips, connecting sensors, processors, and functional modules, is critical to achieving adaptive transformation of diverse data representation formats. This paper proposes a novel adaptive and low-power quantization technique and systematically validates its effectiveness from algorithm to hardware module for industrial IoT applications, covering precise navigation for autonomous vehicles and accurate classification utilizing deep neural networks (DNNs). The proposed quantization method merges an adaptive conversion function from floating-point to fixed-point binaries and an adaptive radix-point determination function, ensuring adequate resolution and minimal error loss of the fixed-point inputs to the edge AI modules. The experimental results demonstrate that the quantization error in the proposed quantization technique contributes negligible errors to the navigation solutions of the strapdown inertial navigation system and the DNNs' top-1 and top-5 classification accuracy (on the order of 10 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{-8}$</tex-math></inline-formula> and 10 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{-7}$</tex-math></inline-formula> ). Moreover, a quantization-on-multiplier (QoM) hardware module is designed, synthesized, and routed in accordance with the proposed quantization technique. The simulation results indicate that the QoM's power consumption and area are 0.1 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$mW$</tex-math></inline-formula> and 649.552 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mu m^{2}$</tex-math></inline-formula> , accounting for 5% and 14.15% of its total energy consumption and area, respectively. With our proposed on-chip quantization technique, the time required to quantize the parameters of DNNs is up to 1142 times shorter than with the existing benchmark off-chip quantization approaches.

Solving the discretised neutron diffusion equations using neural networks

AbstractThis paper presents a new approach which uses the tools within artificial intelligence (AI) software libraries as an alternative way of solving partial differential equations (PDEs) that have been discretised using standard numerical methods. In particular, we describe how to represent numerical discretisations arising from the finite volume and finite element methods by pre‐determining the weights of convolutional layers within a neural network. As the weights are defined by the discretisation scheme, no training of the network is required and the solutions obtained are identical (accounting for solver tolerances) to those obtained with standard codes often written in Fortran or C++. We also explain how to implement the Jacobi method and a multigrid solver using the functions available in AI libraries. For the latter, we use a U‐Net architecture which is able to represent a sawtooth multigrid method. A benefit of using AI libraries in this way is that one can exploit their built‐in technologies to enable the same code to run on different computer architectures (such as central processing units, graphics processing units or new‐generation AI processors) without any modification. In this article, we apply the proposed approach to eigenvalue problems in reactor physics where neutron transport is described by diffusion theory. For a fuel assembly benchmark, we demonstrate that the solution obtained from our new approach is the same (accounting for solver tolerances) as that obtained from the same discretisation coded in a standard way using Fortran. We then proceed to solve a reactor core benchmark using the new approach. For both benchmarks we give timings for the neural network implementation run on a CPU and a GPU, and a serial Fortran code run on a CPU.

An ADC-Less RRAM-Based Computing-in-Memory Macro With Binary CNN for Efficient Edge AI

Resistive random-access memory (RRAM) based non-volatile computing-in-memory (nvCIM) has been regarded as a promising solution to enable efficient data-intensive artificial intelligence (AI) applications on resource-limited edge systems. However, existing weighted-current summation-based nvCIM suffers from device non-idealities and significant time, storage, and energy overheads when processing high-precision analog signals. To address these issues, we propose a 3T2R digital nvCIM macro for a fully hardware-implemented binary convolutional neural network (HBCNN), focusing on accelerating edge AI applications at low weight precision. By quantizing the voltage-division results of RRAMs through inverters, the 3T2R macro provides a stable rail-to-rail output without analog-to-digital converters or sensing amplifiers. Moreover, both batch normalization and sign activation are integrated on-chip. The hybrid simulation results show that the proposed 3T2R digital macro achieves an 86.2% (95.6%) accuracy on the CIFAR-10 (MNIST) dataset, corresponding to a 4.7% (1.9%) accuracy loss compared to the software baselines, which also feature a peak energy efficiency of 51.3 TOPS/W and a minimum latency of 8 ns, realizing an energyefficient, low-latency, and robust AI processor.

Ultradense One-Memristor Ternary-Content-Addressable Memory Based on Ferroelectric Diodes

In this Letter, for the first time, one-memristor (1M)-based ternary-content-addressable memory (TCAM) with an ultradense 4F2 cell area is proposed on a single reconfigurable TiN/hafnium zirconium oxide (HZO)/indium gallium zinc oxide (IGZO)/TiN ferroelectric (FE) diode. By modulating the FE-polarization-controlled Schottky junction that exists at the interface, reconfigured P–N, N–P, and ohmic like junctions of the FE diodes were designed for ‘0’, ‘1’, and ‘X’ storage states in the TCAM cell, respectively. In addition to non-volatile FE polarization for data storage, ambipolar-like behavior induced by the symmetrical junction characteristics was obtained on the diodes for searching queries. Using both non-volatile and ambipolar-like characteristics, typical TCAM functions with a maximum driving on/off ratio of ~500 were confirmed experimentally on a single FE diode, indicating the great potential of this memory device in area-efficient artificial intelligence processors.

A Review on Artificial Intelligence Chip

As chipmakers design different types of chips to enable Artificial Intelligence (AI) applications, the adoption of AI chips has increased recently. To support applications based on deep learning, AI chips have inbuilt AI acceleration and are created with a specialized architecture. One of the key drivers boosting the market's expansion is the increasing integration of AI processors in data centers. The major significance of using AI chips when compared with traditional ICs are fast computational integration and large bandwidth. This study summarizes the need of the AI chips and its functionalities and how the AI chips varies from the general ICs. Finally, a discussion on the potential AI chip initiatives are provided.

A Review on Artificial Intelligence Chip

As chipmakers design different types of chips to enable Artificial Intelligence (AI) applications, the adoption of AI chips has increased recently. To support applications based on deep learning, AI chips have inbuilt AI acceleration and are created with a specialized architecture. One of the key drivers boosting the market's expansion is the increasing integration of AI processors in data centers. The major significance of using AI chips when compared with traditional ICs are fast computational integration and large bandwidth. This study summarizes the need of the AI chips and its functionalities and how the AI chips varies from the general ICs. Finally, a discussion on the potential AI chip initiatives are provided.

Study on the Evaluation of Lung Cancer Patients from the Three Aspects of Emotion.

In recent years, epidemiological survey data have shown that lung cancer is the tumor with the fastest increase in cancer incidence and mortality in China. The incidence and mortality of lung cancer in China rank first among tumors, and 80% of patients die within one year of diagnosis. This paper aims to study the evaluation of lung cancer patients from three aspects of emotion, coping style, and Quality of Life (QoL), expounding on the changes in emotion, coping style, and QoL in lung cancer patients after chemotherapy. We assess the negative emotions and survival of lung cancer patients after chemotherapy quality satisfaction survey research. We investigate the general data, QoL, and coping styles of 219 lung cancer patients undergoing chemotherapy before, during, and after chemotherapy based on the artificial intelligence processor. All survey data are input into SPSS 19 for descriptive and relevant statistical analysis. The experimental results show that under a survey of 219 lung cancer chemotherapy patients with negative emotions and QoL satisfaction after chemotherapy, at a significance level of = 0.05, there is a linear regression relationship between mental resilience and anxiety and depression.

Analysis of the Influence of Network Continuous Care on the Quality of Life of Patients with Coronary Artery Disease (CAD) after PIC.

Studies have shown that most patients after PCI cannot adhere to the cardiac rehabilitation program. The survey found that due to the lack of secondary prevention of cardiac rehabilitation, the phenomenon of drug reduction and withdrawal after PCI is very common after discharge, leading to recurrence of the patient's disease or worsen and repeated hospitalizations, so continuity of care is very important. In this paper, in addition to proposing a network care continuum with artificial intelligence handler in order to improve the healthcare system and provide new ideas for improving the postoperative recovery of CHD patients, we analyze the impact of PCI on the coronary heart disease patients' PCI postoperative quality of life. In the method part, this article introduces the concepts of continuation care and PCI after surgery, introduces the marker delivery algorithm in the field of artificial intelligence, and introduces the SF-36 scale for patient quality of life analysis. This article designs an experiment combining artificial intelligence processors to carry out network continuity care for patients and divides 100 eligible patients into an experimental group and a control group. In the analysis part, the two groups of patients were analyzed in terms of general data comparison, physical function, biochemical indicators, quality of life, and dependence. It can be seen from the experimental analysis that the anxiety and depression of the two groups of subjects have different degrees of decline. The HAMA value of the experimental group is 9.06 ± 0.77, and the HAMD value is 9.18 ± 1.20, which is significantly lower than that of the control group, P < 0.05. It can be seen that the use of network continuation care can reduce the negative emotions of patients more than general care. Through psychological counseling and postoperative follow-up, it can improve the optimism and positive emotions of the patients, reduce the negative emotions of the patients, and improve it to a certain extent.

Obstructive Sleep Apnea Syndrome Treated Using a Positive Pressure Ventilator Based on Artificial Intelligence Processor.

With the acceleration of people's life rhythm, obstructive sleep apnea syndrome appears more and more frequently. This research mainly discusses the treatment of obstructive sleep apnea syndrome with a positive pressure ventilator based on artificial intelligence processor. The information storage function of the smart positive pressure ventilator is included in the local medical terminal, presented after logging in with the user authority. It is mainly composed of data collection, data processing, and medical interface design, which embeds data request, data transmission, data analysis, and detailed tasks such as data compression and storage, and functions such as data display, image drawing, and alarm notification are realized by the medical interface. When the CPAP ventilator transmits respiratory data to the local medical terminal, it sends real-time respiratory information data packets. The data packet is collected and sent in real time in a fixed period and then received and analyzed by the local medical terminal. In the CPAP ventilator telemedicine system, the function of alarm message processing is mainly used to detect the patient's breathing status in real time, extract the alarm-related information, and generate an alarm. This function specifically includes several tasks such as alarm detection, alarm prompt, alarm storage, and remote transmission of alarm messages. The confirmed OSAS patients were pressure-titrated with a smart CPAP ventilator and then treated for 5 hours a day, followed by echocardiography after 5 months of continuous treatment. During the study, the average BMI was (28.9 ± 7.2) kg/m2 and the average AHI index was (53.1 ± 37.8) times/h. This study may help improve the quality of life of patients with obstructive sleep apnea syndrome.

APPLICATION OF METHODS OF THE ARTIFICIAL INTELLIGENCE BY PROCESS PLANNING

The results of the analysis of the current state of the level of automation of process planning at the enterprises of the machine-building industry are presented in the paper. A comparative analysis of two methods of organizing the work of the technological service of the enterprise is carried out, the analysis is carried out on the example of the studied standard product. An enlarged method of calculating the time spent on process planning are presented. The calculations of reducing labor intensity, expressed in monetary terms, are given. The plan of introduction of artificial intelligence in process planning at the enterprises of machine-building industry are presented. The possibilities of further growth of production automation based on computer-aided design systems and artificial intelligence technologies are considered. The analysis of each stage of process planning allows us to conclude that the digitalization of production will allow us to move to a new level of automation, combining individual automation elements into a single integrated system. In conclusion, performance indicators from the introduction of artificial intelligence technologies, increased productivity and reduced labor intensity are noted.

A Multi-Core Controller for an Embedded AI System Supporting Parallel Recognition.

Recent advances in artificial intelligence (AI) technology encourage the adoption of AI systems for various applications. In most deployments, AI-based computing systems adopt the architecture in which the central server processes most of the data. This characteristic makes the system use a high amount of network bandwidth and can cause security issues. In order to overcome these issues, a new AI model called federated learning was presented. Federated learning adopts an architecture in which the clients take care of data training and transmit only the trained result to the central server. As the data training from the client abstracts and reduces the original data, the system operates with reduced network resources and reinforced data security. A system with federated learning supports a variety of client systems. To build an AI system with resource-limited client systems, composing the client system with multiple embedded AI processors is valid. For realizing the system with this architecture, introducing a controller to arbitrate and utilize the AI processors becomes a stringent requirement. In this paper, we propose an embedded AI system for federated learning that can be composed flexibly with the AI core depending on the application. In order to realize the proposed system, we designed a controller for multiple AI cores and implemented it on a field-programmable gate array (FPGA). The operation of the designed controller was verified through image and speech applications, and the performance was verified through a simulator.

Hunch Mining

Intuition and hunches are important tools for experts who make time-critical highly complex decisions in turbulent environments. However, hunches are also elusive and exist below the surface when not being used for immediate decision making. These latent hunches are valuable and can lead to creative solutions outside of a crisis. This paper uses a demonstration experiment to show how hunches can be “mined” using revealed causal mapping techniques. Fifty chief information officers and 88 programmer/analysts were interviewed during a very turbulent time in their organizations. The hunch mining results indicate that hunches can be found throughout experts' communications and can serve as valuable insights into strategic and tactical decision making. This research serves as the first manual step in developing natural language processing artificial intelligence tools for automated hunch mining in expert communications.

Editorial Comment: On the Possibilities of AI, Performance Studies, and Digital Cohabitation

Editorial Comment:On the Possibilities of AI, Performance Studies, and Digital Cohabitation Sean Metzger Over the three years that I initially conceived and eventually edited this special issue, I have become convinced of the pressing need to think through the intersections of performance studies and artificial intelligence (AI). Many factors shape this belief, but they might be summarized in a relatively succinct formulation: Code is performative, and interfaces are theatrical; the elaborations of these claims reveal the shifting of both human and machine ontologies as the networks we inhabit proliferate. To a large extent, the constituent essays in this volume complicate and extend such ideas. The following editorial grounds these elaborations and provides one pathway for viewing their collective charge. Scholars of digital media have established the performativity of computer code.1 For example, Lisa Nakamura has pointed to the performative acts that occur as users create avatars in virtual environments.2 N. Katherine Hayles has explored performativity to account for how computational theory might shift the linguistic framework of this concept provided by J. L. Austin: Computational theory treats computer languages as if they were, in Austin's terms, performative utterances. Although material changes do take place when computers process code (magnetic polarities are changed on a disk), it is the act of attaching significance to these physical changes that constitutes computation as such. … Whereas in performative utterances saying is doing because the action performed is symbolic in nature and does not require physical action in the world, at the basic level of computation doing is saying because physical actions also have a symbolic dimension that corresponds directly to computation.3 From a related perspective, Wendy Chun refers to code as "an inhumanly perfect 'performative' uttered by no one. Unlike any other law or performative utterance, code almost always does what it says because it needs no human acknowledgment."4 Code within this context generally denotes steps or commands for a computer to execute. Coding pertains to, but is not limited to, AI, which relies on sophisticated algorithms. A high-performance algorithm generally refers to any complex formula that "optimally organizes the intake of input and the production of output."5 AI's processes of decision-making often remain unknown, a black box. The ways such algorithms [End Page xiii] increasingly structure our lives have become ever more vexing, leading to what Safiya Noble has called "algorithmic oppression" as "fundamental to the operating system of the web" in her writing on search engines.6 In Noble's research, the results of algorithmic search processes yielded disturbing presentations of bodies for consumption; put otherwise, the histrionic display of Black women on porn sites substituted one vision of femininity for the one that Noble sought (specifically, material that might be interesting to some of the youth in her family). Typing "black girls" in a search engine in 2010 produced unexpected iterations of ostensibly Black womanhood. This example demonstrates both the performative nature of the algorithm and also points to the theatricality of the interface. Broadly speaking, discussions about computers often use "screen" as shorthand for the specific structure called the graphic user interface (GUI). However, as Johanna Drucker has explained, such usage eschews the complexities of human–computer interaction (HCI) that involve keyboards and the like used for "remediating common tasks as formal instructions coded into a plan of menus or buttons."7 Automated interfaces involve aural, visual, and tactile sensory modalities.8 As technology advanced, "graphical icons … no longer just looked like objects; they could also mimic the behavior of the things they resembled."9 Many implications follow from this statement, but Drucker offers one illustration relevant to theatricality in particular: "The GUI is a mediating scrim, a boundary space in which we interact with an abstraction of computation … neither the file folders nor the other objects in the screen space are things; they are icons that represent behaviors and actions we want to perform … and they interpellate a user through disciplinary and scopic regimes."10 She further explains that the GUI effects a "constitutive exchange" that involves "cognitive adaptation and change" through which the interface facilitates "individual and collective subject formation."11 Drucker does not mention theatre studies or...

Special issue on SoC and AI processors

Special issue on SoC and AI processors