Scalable and compact photonic neural chip with low learning-capability-loss

Abstract Photonic computation has garnered huge attention due to its great potential to accelerate artificial neural network tasks at much higher clock rate to digital electronic alternatives. Especially, reconfigurable photonic processor consisting of Mach–Zehnder interferometer (MZI) mesh is promising for photonic matrix multiplier. It is desired to implement high-radix MZI mesh to boost the computation capability. Conventionally, three cascaded MZI meshes (two universal N × N unitary MZI mesh and one diagonal MZI mesh) are needed to express N × N weight matrix with O(N 2) MZIs requirements, which limits scalability seriously. Here, we propose a photonic matrix architecture using the real-part of one nonuniversal N × N unitary MZI mesh to represent the real-value matrix. In the applications like photonic neural network, it probable reduces the required MZIs to O(Nlog2 N) level while pay low cost on learning capability loss. Experimentally, we implement a 4 × 4 photonic neural chip and benchmark its performance in convolutional neural network for handwriting recognition task. Low learning-capability-loss is observed in our 4 × 4 chip compared to its counterpart based on conventional architecture using O(N 2) MZIs. While regarding the optical loss, chip size, power consumption, encoding error, our architecture exhibits all-round superiority.

Download Full-text

A Low-Cost Hardware-Friendly Spiking Neural Network Based on Binary Mram Synapses, Accelerated Using In-Memory Computing

Electronics ◽

10.3390/electronics10192441 ◽

2021 ◽

Vol 10 (19) ◽

pp. 2441

Author(s):

Yihao Wang ◽

Danqing Wu ◽

Yu Wang ◽

Xianwu Hu ◽

Zizhao Ma ◽

...

Keyword(s):

Neural Network ◽

Fixed Point ◽

Low Cost ◽

Random Access ◽

Recognition Task ◽

Learning Rule ◽

Resistive Random Access Memory ◽

Single Step ◽

Spiking Neural Network ◽

Promising Alternative

In recent years, the scaling down that Moore's Law relies on has been gradually slowing down, and the traditional von Neumann architecture has been limiting the improvement of computing power. Thus, neuromorphic in-memory computing hardware has been proposed and is becoming a promising alternative. However, there is still a long way to make it possible, and one of the problems is to provide an efficient, reliable, and achievable neural network for hardware implementation. In this paper, we proposed a two-layer fully connected spiking neural network based on binary MRAM (Magneto-resistive Random Access Memory) synapses with low hardware cost. First, the network used an array of multiple binary MRAM cells to store multi-bit fixed-point weight values. This helps to simplify the read/write circuit. Second, we used different kinds of spike encoders that ensure the sparsity of input spikes, to reduce the complexity of peripheral circuits, such as sense amplifiers. Third, we designed a single-step learning rule, which fit well with the fixed-point binary weights. Fourth, we replaced the traditional exponential Leak-Integrate-Fire (LIF) neuron model to avoid the massive cost of exponential circuits. The simulation results showed that, compared to other similar works, our SNN with 1,184 neurons and 313,600 synapses achieved an accuracy of up to 90.6% in the MNIST recognition task with full-resolution (28 × 28) and full-bit-depth (8-bit) images. In the case of low-resolution (16 × 16) and black-white (1-bit) images, the smaller version of our network with 384 neurons and 32,768 synapses still maintained an accuracy of about 77%, extending its application to ultra-low-cost situations. Both versions need less than 30,000 samples to reach convergence, which is a >50% reduction compared to other similar networks. As for robustness, it is immune to the fluctuation of MRAM cell resistance.

Download Full-text

Small Scaled Autonomous Vehicle

International Journal of Innovative Technology and Exploring Engineering - Special Issue ◽

10.35940/ijitee.k1165.09811s19 ◽

2019 ◽

Vol 8 (11S) ◽

pp. 911-914

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Real Time ◽

Low Cost ◽

Autonomous Vehicle ◽

High Accuracy ◽

The Real ◽

Output Layer ◽

Input Layer ◽

Video Feed

This paper presents a self-driving car using low cost convolutional neural network based platform. The project replicates the real autonomous vehicle/car into a small scaled remote controlled (RC) car. The CNN (convolutional neural network) provides a high accuracy prediction. A small camera is mounted on the RC car for capturing real time video feed to firstly train and then to predict the direction. The camera provides video feed to the input layer of the CNN that has hidden layers, that analyzes and compute the data to predict the output from an output layer that tells rc car to move in predicted direction

Download Full-text

Radiation synovectomy of the knee joint

Nuklearmedizin ◽

10.1055/s-0038-1623935 ◽

2006 ◽

Vol 45 (01) ◽

pp. 57-61

Author(s):

M. Puille ◽

D. Steiner ◽

R. Bauer ◽

R. Klett

Keyword(s):

Knee Joint ◽

Lymph Nodes ◽

Geometric Mean ◽

Radiation Synovectomy ◽

Real Activity ◽

Anterior View ◽

Inguinal Lymph Nodes ◽

The Real ◽

Therapeutic Modalities ◽

Real Value

Summary Aim: Multiple procedures for the quantification of activity leakage in radiation synovectomy of the knee joint have been described in the literature. We compared these procedures considering the real conditions of dispersion and absorption using a corpse phantom. Methods: We simulated different distributions of the activity in the knee joint and a different extra-articular spread into the inguinal lymph nodes. The activity was measured with a gammacamera. Activity leakage was calculated by measuring the retention in the knee joint only using an anterior view, using the geometric mean of anterior and posterior views, or using the sum of anterior and posterior views. The same procedures were used to quantify the activity leakage by measuring the activity spread into the inguinal lymph nodes. In addition, the influence of scattered rays was evaluated. Results: For several procedures we found an excellent association with the real activity leakage, shown by an r² between 0.97 and 0.98. When the real value of the leakage is needed, e. g. in dosimetric studies, simultaneously measuring of knee activity and activity in the inguinal lymph nodes in anterior and posterior views and calculation of the geometric mean with exclusion of the scatter rays was found to be the procedure of choice. Conclusion: When measuring of activity leakage is used for dosimetric calculations, the above-described procedure should be used. When the real value of the leakage is not necessary, e. g. for comparing different therapeutic modalities, several of the procedures can be considered as being equivalent.

Download Full-text

Low-Cost Hardware Implementation of Human Blood Identification Device Using Feed-Forward Artificial Neural Network on FPGA

10.31227/osf.io/dp63x ◽

2018 ◽

Author(s):

Rizki Eka Putri ◽

Denny Darlis

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Low Cost ◽

Image Resolution ◽

Blood Type ◽

Feed Forward ◽

Testing Facility ◽

System Input ◽

Artificial Neural ◽

Short Time

This article was under review for ICELTICS 2018 -- In the medical world there is still service dissatisfaction caused by lack of blood type testing facility. If the number of tested blood arise, a lot of problems will occur so that electronic devices are needed to determine the blood type accurately and in short time. In this research we implemented an Artificial Neural Network on Xilinx Spartan 3S1000 Field Programable Gate Array using XSA-3S Board to identify the blood type. This research uses blood sample image as system input. VHSIC Hardware Discription Language is the language to describe the algorithm. The algorithm used is feed-forward propagation of backpropagation neural network. There are 3 layers used in design, they are input, hidden1, and output. At hidden1layer has two neurons. In this study the accuracy of detection obtained are 92%, 92%, 92%, 90% and 86% for 32x32, 48x48, 64x64, 80x80, and 96x96 pixel blood image resolution, respectively.

Download Full-text

The 'Real' Value of Conflicted Analyst Coverage

SSRN Electronic Journal ◽

10.2139/ssrn.1571418 ◽

2012 ◽

Author(s):

Stacey E. Jacobsen ◽

Irina Stefanescu ◽

Xiaoyun Yu

Keyword(s):

Analyst Coverage ◽

The Real ◽

Real Value

Download Full-text

Real-Time Low-Cost Drift Compensation for Chemical Sensors Using a Deep Neural Network with Hadamard Transform and Additive Layers

IEEE Sensors Journal ◽

10.1109/jsen.2021.3084220 ◽

2021 ◽

pp. 1-1

Author(s):

Diaa Badawi ◽

Agamyrat Agambayev ◽

Sule Ozev ◽

A. Enis Cetin

Keyword(s):

Neural Network ◽

Real Time ◽

Chemical Sensors ◽

Deep Neural Network ◽

Low Cost ◽

Hadamard Transform ◽

Drift Compensation

Download Full-text

Recurrent Neural Network for Human Activity Recognition in Embedded Systems Using PPG and Accelerometer Data

Electronics ◽

10.3390/electronics10141715 ◽

2021 ◽

Vol 10 (14) ◽

pp. 1715

Author(s):

Michele Alessandrini ◽

Giorgio Biagetti ◽

Paolo Crippa ◽

Laura Falaschetti ◽

Claudio Turchetti

Keyword(s):

Neural Network ◽

Embedded System ◽

Activity Recognition ◽

Human Activity ◽

Recurrent Neural Network ◽

Low Cost ◽

Low Complexity ◽

Human Activity Recognition ◽

Motion Artifacts ◽

Accelerometer Data

Photoplethysmography (PPG) is a common and practical technique to detect human activity and other physiological parameters and is commonly implemented in wearable devices. However, the PPG signal is often severely corrupted by motion artifacts. The aim of this paper is to address the human activity recognition (HAR) task directly on the device, implementing a recurrent neural network (RNN) in a low cost, low power microcontroller, ensuring the required performance in terms of accuracy and low complexity. To reach this goal, (i) we first develop an RNN, which integrates PPG and tri-axial accelerometer data, where these data can be used to compensate motion artifacts in PPG in order to accurately detect human activity; (ii) then, we port the RNN to an embedded device, Cloud-JAM L4, based on an STM32 microcontroller, optimizing it to maintain an accuracy of over 95% while requiring modest computational power and memory resources. The experimental results show that such a system can be effectively implemented on a constrained-resource system, allowing the design of a fully autonomous wearable embedded system for human activity recognition and logging.

Download Full-text

End-to-end convolutional neural network enables COVID-19 detection from breath and cough audio: a pilot study

BMJ Innovations ◽

10.1136/bmjinnov-2021-000668 ◽

2021 ◽

Vol 7 (2) ◽

pp. 356-362

Author(s):

Harry Coppock ◽

Alex Gaskell ◽

Panagiotis Tzirakis ◽

Alice Baird ◽

Lyn Jones ◽

...

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Low Cost ◽

Area Under The Curve ◽

Alternative Form ◽

Economic Damage ◽

Audio Signals ◽

Operating Characteristics ◽

Data Set ◽

Empirical Performance

BackgroundSince the emergence of COVID-19 in December 2019, multidisciplinary research teams have wrestled with how best to control the pandemic in light of its considerable physical, psychological and economic damage. Mass testing has been advocated as a potential remedy; however, mass testing using physical tests is a costly and hard-to-scale solution.MethodsThis study demonstrates the feasibility of an alternative form of COVID-19 detection, harnessing digital technology through the use of audio biomarkers and deep learning. Specifically, we show that a deep neural network based model can be trained to detect symptomatic and asymptomatic COVID-19 cases using breath and cough audio recordings.ResultsOur model, a custom convolutional neural network, demonstrates strong empirical performance on a data set consisting of 355 crowdsourced participants, achieving an area under the curve of the receiver operating characteristics of 0.846 on the task of COVID-19 classification.ConclusionThis study offers a proof of concept for diagnosing COVID-19 using cough and breath audio signals and motivates a comprehensive follow-up research study on a wider data sample, given the evident advantages of a low-cost, highly scalable digital COVID-19 diagnostic tool.

Download Full-text