Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170404
V. Srimaheswaran, N. Niveditha, M. Rajan Singaravel
A high voltage gain non-isolated DC-DC converter is proposed in this work using two active switches and six passive switches. The proposed converter has Switched Inductor, Capacitor network (SLC) and voltage quadrupler (VQ) network at the output side, which boost the voltage at the load. The proposed converter (SLCVQ) has higher voltage gain with a lower duty ratio and less components compared to the existing converters. The SLCVQ converter topology is presented and the comprehensive operation of the circuit is discussed with mathematical calculations. The components of the proposed converter are designed for 500 W, 50kHz, 650V output voltage and simulated using a PSIM environment. The simulation results confirmed that the proposed converter has high voltage gain, which can be used for renewable energy sources applications like smart grids, microgrids, etc.
{"title":"Design and Analysis of High Gain Converter with Low Duty-Ratio for DC Grid","authors":"V. Srimaheswaran, N. Niveditha, M. Rajan Singaravel","doi":"10.1109/IConSCEPT57958.2023.10170404","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170404","url":null,"abstract":"A high voltage gain non-isolated DC-DC converter is proposed in this work using two active switches and six passive switches. The proposed converter has Switched Inductor, Capacitor network (SLC) and voltage quadrupler (VQ) network at the output side, which boost the voltage at the load. The proposed converter (SLCVQ) has higher voltage gain with a lower duty ratio and less components compared to the existing converters. The SLCVQ converter topology is presented and the comprehensive operation of the circuit is discussed with mathematical calculations. The components of the proposed converter are designed for 500 W, 50kHz, 650V output voltage and simulated using a PSIM environment. The simulation results confirmed that the proposed converter has high voltage gain, which can be used for renewable energy sources applications like smart grids, microgrids, etc.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123488907","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
AI Based Virtual Mouse is an AI-based project that allows users to access the mouse with hand gestures, without having to physically touch the mouse. This solution integrates a chatbot with a basic camera, rather than a traditional mouse, to handle mouse operations using user-friendly voice instructions. This solution cuts down on the need for hardware such as a wireless mouse or a Bluetooth mouse by capturing hand motions and fingertip recognition using computer vision and a webcam or built-in camera. The user can perform various actions such as clicking, left-clicking, right-clicking, and dragging with different hand gestures. This system necessitates the use of a webcam, microphone, and speaker.The camera’s output will be shown on the screen with the intention that the user may fine-tune it. As a result, the suggested solution eliminates the requirement for human involvement and the computer’s reliance on external devices.
{"title":"Jarvis - AI Based Virtual Mouse","authors":"Aditi Khandagale, Nidhi Thakkar, Swarali Patil, Vaibhavi Jadhav, Charusheela Nehete","doi":"10.1109/IConSCEPT57958.2023.10170722","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170722","url":null,"abstract":"AI Based Virtual Mouse is an AI-based project that allows users to access the mouse with hand gestures, without having to physically touch the mouse. This solution integrates a chatbot with a basic camera, rather than a traditional mouse, to handle mouse operations using user-friendly voice instructions. This solution cuts down on the need for hardware such as a wireless mouse or a Bluetooth mouse by capturing hand motions and fingertip recognition using computer vision and a webcam or built-in camera. The user can perform various actions such as clicking, left-clicking, right-clicking, and dragging with different hand gestures. This system necessitates the use of a webcam, microphone, and speaker.The camera’s output will be shown on the screen with the intention that the user may fine-tune it. As a result, the suggested solution eliminates the requirement for human involvement and the computer’s reliance on external devices.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115280230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170642
P. Ramanathan, M. Ericsson
Pathogenic bacterial growth detection and monitoring is an important scientific process in the field of quality control in the food, water, and medical industries. Very-large-scale process of such bacteria growth monitoring is possible only with an automated process. The mechanism must make sure that the sample is continuously monitored, and detected, data is communicated to supervisors and managers, and data is stored historically retrievable for quality control and analysis. A manual bacteria inspection among the Petri dishes incubated of such bacterial growth in food processing was attempted for automation. The manual inspection in a microbiological industry involves; an operator inspecting the input petri discs to check if there are bacteria, writing down the barcode of the corresponding petri dish, and then sorting the Petri discs depending on the bacterial growth. In this automation attempt of automatizing this petri-disc inspection, the project was split into two phases. 1. Building a vision system to detect bacteria, developing of an algorithm to quantify the growth, and registering the barcode in a registry. 2. The second phase is to design a robot system with programming and define the layout of the station. The development of an intelligent robotized machine vision automated system proves the concept of a major industrial practice that has the potential to significantly increase the quality and productivity of bacterial growth, with increased throughput.
{"title":"Development of an Intelligent Robotized Machine Vision Automated System for Bacterial Growth Monitoring","authors":"P. Ramanathan, M. Ericsson","doi":"10.1109/IConSCEPT57958.2023.10170642","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170642","url":null,"abstract":"Pathogenic bacterial growth detection and monitoring is an important scientific process in the field of quality control in the food, water, and medical industries. Very-large-scale process of such bacteria growth monitoring is possible only with an automated process. The mechanism must make sure that the sample is continuously monitored, and detected, data is communicated to supervisors and managers, and data is stored historically retrievable for quality control and analysis. A manual bacteria inspection among the Petri dishes incubated of such bacterial growth in food processing was attempted for automation. The manual inspection in a microbiological industry involves; an operator inspecting the input petri discs to check if there are bacteria, writing down the barcode of the corresponding petri dish, and then sorting the Petri discs depending on the bacterial growth. In this automation attempt of automatizing this petri-disc inspection, the project was split into two phases. 1. Building a vision system to detect bacteria, developing of an algorithm to quantify the growth, and registering the barcode in a registry. 2. The second phase is to design a robot system with programming and define the layout of the station. The development of an intelligent robotized machine vision automated system proves the concept of a major industrial practice that has the potential to significantly increase the quality and productivity of bacterial growth, with increased throughput.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124851195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170038
G. Ashwini, T. Ramashri
The onset of COVID-19 pandemic had a great impact on the health, economy and livelihood of so many lives around the world. Early identification of positive cases and isolation followed by treatment is very crucial in order to receive prompt treatment and prevent the virus from spreading further. Chest X-ray (CXR) and Computed Tomography (CT) are widespread and cost-effective medical imaging radiographic tools which are presently used for diagnosing the covid-19 quickly. A crucial step towards establishing a quick COVID-19 pre-diagnosis and reducing the workload on medical professionals is the use of deep learning algorithms to identify positive CXR and CT pictures of infected individuals. However, the CXR images have complicated edge structures and rich texture details that are sensitive to noise, which can interfere with the machines and doctors diagnosis. This paper presents two state-of-art denoising techniques, Noise2Noise(N2N) and Noise2Void(N2V),to eliminate the noises that were added to COVID-19 chest x-ray scan image and CT medical image modalities by additive Gaussian noise, Speckle noise, and salt-and-pepper noise. These two techniques do not require a pair of noisy and clean photos; instead, they denoise a single noisy image. Based on the study, Noise2Void performs well in removing Gaussian, Speckle, and salt-and-pepper noise from CXR image modalities. Similarly, Noise2Noise performance is good to remove only Gaussian noise in CT images. In the case of Speckle and salt and pepper noise in CT images. Noise2void gives better quality images with better PSNR and SSIM. The results are measured quantitatively and qualitatively using Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Measure(SSIM). In this paper, we experiment two learning based methods to de-noise images with high noise. The proposed method is beneficial for applications involving images that are highly prone to noise.
{"title":"Denoising of COVID-19 CT and chest X-ray images using deep learning techniques for various noises using single image","authors":"G. Ashwini, T. Ramashri","doi":"10.1109/IConSCEPT57958.2023.10170038","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170038","url":null,"abstract":"The onset of COVID-19 pandemic had a great impact on the health, economy and livelihood of so many lives around the world. Early identification of positive cases and isolation followed by treatment is very crucial in order to receive prompt treatment and prevent the virus from spreading further. Chest X-ray (CXR) and Computed Tomography (CT) are widespread and cost-effective medical imaging radiographic tools which are presently used for diagnosing the covid-19 quickly. A crucial step towards establishing a quick COVID-19 pre-diagnosis and reducing the workload on medical professionals is the use of deep learning algorithms to identify positive CXR and CT pictures of infected individuals. However, the CXR images have complicated edge structures and rich texture details that are sensitive to noise, which can interfere with the machines and doctors diagnosis. This paper presents two state-of-art denoising techniques, Noise2Noise(N2N) and Noise2Void(N2V),to eliminate the noises that were added to COVID-19 chest x-ray scan image and CT medical image modalities by additive Gaussian noise, Speckle noise, and salt-and-pepper noise. These two techniques do not require a pair of noisy and clean photos; instead, they denoise a single noisy image. Based on the study, Noise2Void performs well in removing Gaussian, Speckle, and salt-and-pepper noise from CXR image modalities. Similarly, Noise2Noise performance is good to remove only Gaussian noise in CT images. In the case of Speckle and salt and pepper noise in CT images. Noise2void gives better quality images with better PSNR and SSIM. The results are measured quantitatively and qualitatively using Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Measure(SSIM). In this paper, we experiment two learning based methods to de-noise images with high noise. The proposed method is beneficial for applications involving images that are highly prone to noise.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128318392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170131
A. Preethi, P. Dhanalakshmi
Digital video is more prevalent nowadays because of more usage of video data among users. The short and catchy videos among social media attract the attention of people. On the same time, the lengthy videos are found to be left without being fully watched. So, video captioning overcomes this issue by automatically generating captions for a video. The process of generating meaningful natural language sentences for the corresponding scenes in the video is called video captioning. Video captioning involves two steps, namely, feature extraction and caption generation. Here, the pre-trained CNN such as InceptionV3 and VGG16 were used for extracting the features from the video. The caption generation is done through LSTM with the help of extracted features. The relevant captions are achieved using LSTM with the help of word embeddings.
{"title":"Video Captioning using Pre-Trained CNN and LSTM","authors":"A. Preethi, P. Dhanalakshmi","doi":"10.1109/IConSCEPT57958.2023.10170131","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170131","url":null,"abstract":"Digital video is more prevalent nowadays because of more usage of video data among users. The short and catchy videos among social media attract the attention of people. On the same time, the lengthy videos are found to be left without being fully watched. So, video captioning overcomes this issue by automatically generating captions for a video. The process of generating meaningful natural language sentences for the corresponding scenes in the video is called video captioning. Video captioning involves two steps, namely, feature extraction and caption generation. Here, the pre-trained CNN such as InceptionV3 and VGG16 were used for extracting the features from the video. The caption generation is done through LSTM with the help of extracted features. The relevant captions are achieved using LSTM with the help of word embeddings.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124488620","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170616
Naga Sravanthi Puppala, R. Manoharan
Blockchain mining has received a growing amount of interest as the modern technologies started using Blockchain network as it provides immutability and secure access through consensus process. In Blockchain networks that rely on proof-of-work (POW) for consensus process, where miners compete to solve crypto-puzzles and publish new blocks in order to gain rewards. Because solo mining is challenging, most miners decide to join a mining pool. Given the variety of mining pools and the potential adoption of different reward systems, many questions remain open regarding the miner’s selection in pool mining. However, ensuring trust of a node during consensus and mining process has few challenges to explore. Nodes in a mining pool contribute their processing power toward the effort of adding a block as part of consensus. If the pool is successful in these efforts, they receive a reward. Therefore, to ensure the execution of the consensus protocol a trust model is proposed in our work. Our proposed trust paradigm focused on firstly, how to evaluate the unstable behavior of miners based on their performance on the network. Secondly, how to identify highly ranked-trusted miner in mining pool to maximize pool profit. These issues necessitate ranking methods in mining pool, to rank reputed miners. Therefore, this paper proposes a trust model to identify highly ranked-trusted miners in pool mining. Further, proposed Trust Model is suitably analyzed for transaction volume, latency, and block propagation time with the Hyper ledger framework to ensure robustness.
{"title":"Trust Model to Identify Reputed Miners in Blockchain Pool Mining","authors":"Naga Sravanthi Puppala, R. Manoharan","doi":"10.1109/IConSCEPT57958.2023.10170616","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170616","url":null,"abstract":"Blockchain mining has received a growing amount of interest as the modern technologies started using Blockchain network as it provides immutability and secure access through consensus process. In Blockchain networks that rely on proof-of-work (POW) for consensus process, where miners compete to solve crypto-puzzles and publish new blocks in order to gain rewards. Because solo mining is challenging, most miners decide to join a mining pool. Given the variety of mining pools and the potential adoption of different reward systems, many questions remain open regarding the miner’s selection in pool mining. However, ensuring trust of a node during consensus and mining process has few challenges to explore. Nodes in a mining pool contribute their processing power toward the effort of adding a block as part of consensus. If the pool is successful in these efforts, they receive a reward. Therefore, to ensure the execution of the consensus protocol a trust model is proposed in our work. Our proposed trust paradigm focused on firstly, how to evaluate the unstable behavior of miners based on their performance on the network. Secondly, how to identify highly ranked-trusted miner in mining pool to maximize pool profit. These issues necessitate ranking methods in mining pool, to rank reputed miners. Therefore, this paper proposes a trust model to identify highly ranked-trusted miners in pool mining. Further, proposed Trust Model is suitably analyzed for transaction volume, latency, and block propagation time with the Hyper ledger framework to ensure robustness.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126794897","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170228
K. Adarsh Geoffrey Daniel, Bertia Albert
In a technology filled world with a lot of online data it is very hard to find a person’s attitude or behavior or his/her likeness. This project, which can predict the likeness of the person using their online logs can be used for this. The present study of a person is based on their online activities, but to identify which category they like the most comes from their personal behavior on the domains they visit. This particular project finds the likeness of a person based on their most liked webpages. This uses the records of DNS logs and tries to identify the most seen webpages and figures out which category they like the most. The program used in this project is a multiclass classification model that would classify and predict the type of webpage the user has visited the most. This will help in effectively predicting the particular person’s likeness. With this method tests were conducted with three main algorithms, Support Vector Machine, Convolutional Neural Network and Naive Bayes out of which we were able to get an accuracy of 95% using the Naive Bayes algorithm, which helped in predicting the user’s likeness. This can further be enhanced with much higher real time log activity finder and real time log analyser which helps in finding or keeping a track of the person’s behavior. This program can widely be used to study humans.
{"title":"Analyzing the likeness of a person based on DNS logs using machine learning","authors":"K. Adarsh Geoffrey Daniel, Bertia Albert","doi":"10.1109/IConSCEPT57958.2023.10170228","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170228","url":null,"abstract":"In a technology filled world with a lot of online data it is very hard to find a person’s attitude or behavior or his/her likeness. This project, which can predict the likeness of the person using their online logs can be used for this. The present study of a person is based on their online activities, but to identify which category they like the most comes from their personal behavior on the domains they visit. This particular project finds the likeness of a person based on their most liked webpages. This uses the records of DNS logs and tries to identify the most seen webpages and figures out which category they like the most. The program used in this project is a multiclass classification model that would classify and predict the type of webpage the user has visited the most. This will help in effectively predicting the particular person’s likeness. With this method tests were conducted with three main algorithms, Support Vector Machine, Convolutional Neural Network and Naive Bayes out of which we were able to get an accuracy of 95% using the Naive Bayes algorithm, which helped in predicting the user’s likeness. This can further be enhanced with much higher real time log activity finder and real time log analyser which helps in finding or keeping a track of the person’s behavior. This program can widely be used to study humans.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126730774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170384
A. Chaudhari, Sharvit Shewade, Tinish Uge, Awanti Thigale, Prathmesh T. Talekar, Sidhhesh Shirsath
Air quality monitoring is an important tool for improving air quality, protecting public health, and ensuring compliance with regulations. It can also be used to identify pollution sources, monitor climate change, or support research and development. The goal is to identify the most significant sources and amounts of air pollution in truck compartments, with a focus on particles. Along with investigating how the design of the truck’s ventilation system and air purification equipment affects air quality, and developing a robust and effective methodology for measurement in traffic environments. Levels of air pollution were measured both inside and outside the truck cabin, as well as comfort parameters while driving in real traffic, during five extensive full-scale and a number of smaller, scaled-down measurement campaigns. The majority of the experiment was conducted in a truck cabin, which included a dusty environment, city driving, and a stationary road. LPG or butane concentrations of size resolved particles are measured using physical components. Simultaneously, the driver is alerted by calculating vital health characteristics such as heart rate, oxygen level, and so on. After experimentation with a real truck the project was able monitor the air quality and altering the driver by delivering the message.
{"title":"IOT based air quality detection in truck cabins","authors":"A. Chaudhari, Sharvit Shewade, Tinish Uge, Awanti Thigale, Prathmesh T. Talekar, Sidhhesh Shirsath","doi":"10.1109/IConSCEPT57958.2023.10170384","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170384","url":null,"abstract":"Air quality monitoring is an important tool for improving air quality, protecting public health, and ensuring compliance with regulations. It can also be used to identify pollution sources, monitor climate change, or support research and development. The goal is to identify the most significant sources and amounts of air pollution in truck compartments, with a focus on particles. Along with investigating how the design of the truck’s ventilation system and air purification equipment affects air quality, and developing a robust and effective methodology for measurement in traffic environments. Levels of air pollution were measured both inside and outside the truck cabin, as well as comfort parameters while driving in real traffic, during five extensive full-scale and a number of smaller, scaled-down measurement campaigns. The majority of the experiment was conducted in a truck cabin, which included a dusty environment, city driving, and a stationary road. LPG or butane concentrations of size resolved particles are measured using physical components. Simultaneously, the driver is alerted by calculating vital health characteristics such as heart rate, oxygen level, and so on. After experimentation with a real truck the project was able monitor the air quality and altering the driver by delivering the message.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127069819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170596
Vijaya Kumar, Suresh Balanethiram
Spiking Neural Networks (SNNs) are a promising alternative to traditional Deep Neural Networks (DNNs) due to their ability to operate in low-power event-driven mode. However, training SNNs from scratch remains challenging, and conversion-based SNNs derived from pre-trained DNNs have become popular. In this paper, we focus on generating learnable parameters for the inference phase by analyzing the timing window of rate-coded spiking activation using N-MNIST digit classification. We compare the training accuracy of a non-spiking ANN model with the spike ignored and spike-aware spiking activation models trained at different time intervals. We also use regularization to control the mean spike rate of neurons and include a moving-average pooling layer to improve classification accuracy. We provide insights into optimizing the timing window of rate-coded spiking activation for energy-efficient and accurate SNN inference. Our results show that spike-aware training with regularization and moving-average pooling improves convergence and achieves high accuracy. These findings can help improve the training of SNNs for various AI applications.
{"title":"Spike-Aware Training and Timing Window Optimization for Energy-Efficient Inference in Conversion-Based Spiking Neural Networks","authors":"Vijaya Kumar, Suresh Balanethiram","doi":"10.1109/IConSCEPT57958.2023.10170596","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170596","url":null,"abstract":"Spiking Neural Networks (SNNs) are a promising alternative to traditional Deep Neural Networks (DNNs) due to their ability to operate in low-power event-driven mode. However, training SNNs from scratch remains challenging, and conversion-based SNNs derived from pre-trained DNNs have become popular. In this paper, we focus on generating learnable parameters for the inference phase by analyzing the timing window of rate-coded spiking activation using N-MNIST digit classification. We compare the training accuracy of a non-spiking ANN model with the spike ignored and spike-aware spiking activation models trained at different time intervals. We also use regularization to control the mean spike rate of neurons and include a moving-average pooling layer to improve classification accuracy. We provide insights into optimizing the timing window of rate-coded spiking activation for energy-efficient and accurate SNN inference. Our results show that spike-aware training with regularization and moving-average pooling improves convergence and achieves high accuracy. These findings can help improve the training of SNNs for various AI applications.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134363172","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-25DOI: 10.1109/IConSCEPT57958.2023.10170495
S. Syedakbar, S. Geerthana, S. Nithya
We proposed a two dimensional photonic crystal based 3-bit code converter which converts binary codes to gray codes using hexagonal lattice. Using hexagonal silicon pillars in dual T-shaped waveguides code converter is realized by means of XOR gates. By employing the cascaded layout of XOR logic gates, the output can be arrived by using interference effect. Finite Difference Time Domain methodology is used to design and analyze the performance of the designed code converters. The design renders a contrast ratio of 12.557 dB for the binary to gray code converter with optimized refractive index and silicon rod radius values.
{"title":"Photonic Crystal Based Code Converter-Binary to Gray Code","authors":"S. Syedakbar, S. Geerthana, S. Nithya","doi":"10.1109/IConSCEPT57958.2023.10170495","DOIUrl":"https://doi.org/10.1109/IConSCEPT57958.2023.10170495","url":null,"abstract":"We proposed a two dimensional photonic crystal based 3-bit code converter which converts binary codes to gray codes using hexagonal lattice. Using hexagonal silicon pillars in dual T-shaped waveguides code converter is realized by means of XOR gates. By employing the cascaded layout of XOR logic gates, the output can be arrived by using interference effect. Finite Difference Time Domain methodology is used to design and analyze the performance of the designed code converters. The design renders a contrast ratio of 12.557 dB for the binary to gray code converter with optimized refractive index and silicon rod radius values.","PeriodicalId":240167,"journal":{"name":"2023 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)","volume":"142 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133271939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：