Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
8 results
Search Results
Item Empirical Study on Multi Convolutional Layer-based Convolutional Neural Network Classifier for Plant Leaf Disease Detection(Institute of Electrical and Electronics Engineers Inc., 2020) Sunil, C.K.; Jaidhar, C.D.; Patil, N.Recognizing the plant disease automatically in real-time by examining a plant leaf image is highly essential for farmers. This work focuses on an empirical study on Multi Convolutional Layer-based Convolutional Neural Network (MCLCNN) classifier to measure the detection efficacy of MCLCNN on recognizing plant leaf image as being healthy or diseased. To achieve this, a set of experiments were conducted with three distinct plant leaf datasets. Each of the experiments were conducted by setting kernel size of 3× 3 and each experiment was conducted independently with different epochs i.e., 50, 75, 100, 125, and 150. The MCLCNN classifier achieved minimum accuracy of 87.47% with 50 epochs and maximum accuracy of 99.25% with 150 epochs for the Peach plant leaves. © 2020 IEEE.Item Smart Energy Meter Calibration: An Edge Computation Method: Poster(Association for Computing Machinery, Inc, 2021) Dubara, H.V.; Parihar, M.; Ramamritham, K.Smart meters are the backbone of smart grids. They provide real time electricity consumption data and and are widely used for measuring, monitoring and analyzing energy consumption. Sometimes, they enable users to perform corrective actions. But, to facilitate proper data analysis, it is imperative that data be accurate or have minimum error. This paper presents an edge deployed smart meter error correction algorithm that utilises Clustering (using K-Means algorithm) and Feed-Forward Artificial Neural Networks (ANN). An edge device, a Raspberry Pi Module, connects smart meters to the internet. The algorithm maps (possibly erroneous) readings of our in-house developed meters to readings of calibrated standard off-the-shelf (Schneider) meters. Usage of Clustering with ANN has helped substantially improve the accuracy of the readings from a previously used linear regression designed for the same purpose. An accuracy of 70-75% was achieved while using linear regression, whereas the proposed algorithm obtains accuracy in the range of 84.47-88%. The neural networks are also less complex, making them suitable for deployment in Raspberry Pi 3B based embedded hardware systems. © 2021 ACM.Item Design and Implementation of Reconfigurable Neural Network Accelerator(Institute of Electrical and Electronics Engineers Inc., 2022) Shenoy, M.S.; Ramesh Kini, M.General-purpose CPUs are sluggish and inefficient when used for computationally intensive applications including in neural networks. It is preferable to develop specialized hardware that can do a large number of multiply-accumulate operations rapidly and efficiently to execute such applications. The Re-configurable Neural Network Accelerator (RNNA) architecture that has been designed is appropriate for a variety of neural network applications. The computational resource requirements vary depending on the application; hence, mapping the application to the available set of resources requires reconfigurability. The fundamental unit of the RNNA is composed of a variety of Multiply-Accumulate (MAC) units, registers, and Address Generation Units (AGU). When compared to the computation performed by a single MAC array, the RNNA with four MAC arrays reduces the time required by approximately 75%. On the Nexys4 DDR Artix-7 FPGA board, RNNA was tested and implemented with a clock frequency of up to 60MHz and power consumption of 0.243W. © 2022 IEEE.Item Sentiment Analysis and Homophobia Detection of YouTube Comments(CEUR-WS, 2022) Ugursandi, S.; Anand Kumar, A.M.Sentiment analysis identifies a graded scale of opinions or emotional responses to a particular subject. Many industries and organisations have been actively researching this area for more than 20 years. The key to understand a user’s behaviour while responding on a social media site is to understand their feelings. In contemporary research, a sentence’s content is evaluated, the emotion predicted, that helps researchers gain an insight on the reaction of an individual towards a social media topic. Here, a sentence’s text data is analysed using several Natural Language Processing techniques before being utilised to categorise this multi-class issue. The detection of homophobia and transphobia in comments on YouTube or other social media sites is second objective of this work. Anger, discomfort, or suspicion against Lesbian, Gay, Bisexual and Transgender people is known as homophobia. It can incite individuals to feel panic, dislike, disrespect, aggression, or wrath. By identifying such occurrences on social media, we can better understand how society works and how people behave. The goal of this work is to analyze social media texts such as comments from YouTube and detect homophobic sentiments using deep learning or machine learning models. In this work 6-layer classification model is used, the F1-Score for sentiment identification using the proposed model in this study was 0.5 on multi-class classification and 0.97 on homophobic/transphobic classification and achieved 1st rank on Homophobic detection in Malayalam language and 4th rank for sentiment analysis in Kannada language. © 2022 Copyright for this paper by its authors.Item A Novel Approach towards Windows Malware Detection System Using Deep Neural Networks(Elsevier B.V., 2022) Divakarla, U.; Reddy, K.H.K.; Chandrasekaran, K.Now-a-day's malicious software is increasing in numbers and at present becomes more harmful for any digital equipment like mobile, tablet, and computers. Traditional techniques like static and dynamic analysis, signature-based detection methods are become absolute and not effective at all. The advanced techniques like code encryption and code packing techniques can be used to hide detection; polymorphic malware is a new class of malware that changes their code structure from time to time to avoid detection, so there is a need for an intelligent system which can efficiently analyze the features of a new, unknown executable file and classify it correctly. There have been learning-based malware detection systems proposed in the literature, but most of those proposed approaches present a high accuracy over a small dataset, whereas the performance is very poor over industry-standard datasets. Operating system like windows is always in prime malware target because of the sheer high number of users. This paper proposes a simple, deep learning-based detection approachthat classifies a specified executable into benign or harmful. It has been trained using EMBER, an industry-level Windows malware dataset and tests with an accuracy of 87.76%. © 2023 The Authors. Published by Elsevier B.V.Item CNN-based Soil Fertility Classification with Fertilizer Prescription(Institute of Electrical and Electronics Engineers Inc., 2023) Sujatha, M.; Jaidhar, C.D.Soil fertility plays a vital role in crop growth, and thus, the rapid acquisition of soil fertility levels and applying precise fertilizer is significant for sustainable agricultural development. However, obtaining accurate soil fertility estimates proves difficult due to the traditional practice of laboratory analysis of soil samples. This study proposes fertilizer prescriptions based on the Convolutional Neural Networks (CNNs) classifier results. The soil fertility is classified as HIGH, MEDIUM, or LOW fertile based on the chemical measurements of soil parameters, including EC, pH, OC, P, K, S, Zn, B, Cu, Fe, and Mn. The experiments were carried out by varying kernel size from $3\times 3$ to $7\times 7$ and input grid size from $11\times 11$ to $13\times 13$. The proposed approach outperformed with an Accuracy of 97.24% without oversampling the dataset for kernel size $3\times 3$ and input grid size $11\times 11$. Further, for the dataset oversampled using Synthetic Minority Oversampling (SMOTE) technique, the proposed approach achieved the highest Accuracy of 97.52% for kernel size $3\times 3$ and input grid size $12\times 12$. The study helps in the precise application of fertilizers for specific crops based on classification results. © 2023 IEEE.Item Detection of Pneumonia from Chest X-Ray Images(Institute of Electrical and Electronics Engineers Inc., 2024) Shetty, S.P.; Mamatha, N.; Shetty, M.; Keerthana, S.; Shetty D, P.Pneumonia is a dangerous which is caused by various viral agents. The diagnosis and treatment of pneumonia can be difficult because of the similarities with other lung diseases, which underscores the importance of chest x-rays for an early detection. This work presents techniques of pneumonia detection implementing CNNs, VGG16 and ResNet152V2 architectures, together with the Gradient Descent optimization method. The model is trained and tested on one of Kaggle's dataset which have 5,836 images that are labeled. This system automatically extract features from the chest X-Ray images and uses Gradient Descent optimization to improve its ability to differentiate between the pneumonia patients and healthy cases. For given dataset, the result provides accuracy of 96.56%, 95.34%, 92.9% and 94.23% for RestNet152V2,CNN,VGG16 and Gradient Descent respectively. Therefore this framework will facilitate to the detection of lung disease for experts and doctors as well. © 2024 IEEE.Item Soft computing techniques during drilling of bi-directional carbon fiber reinforced composite(Elsevier Ltd, 2016) Shetty, N.; Herbert, M.A.; Shetty, R.; Shetty, D.S.; Vijay, G.S.Due to the intricacy of machining processes and inconsistency in material properties, analytical models are often unable to describe the mechanics of machining of carbon fiber reinforced polymer (CFRP) composites. Recently, soft computing techniques are used as alternate modeling and analyzing methods, which are usually robust and capable of yielding comprehensive, precise, and unswerving solutions. In this paper, drilling experiments as per the Taguchi L27 experimental layout are carried out on bi-directional carbon fiber reinforced polymer (BD CFRP) composite laminates using three types of drilling tools: high speed steel (HSS), uncoated solid carbide (USC) and titanium nitride coated SC (TiN-SC). The focus of this work is to determine the best drilling tool that produces good quality drilled holes in BD CFRP composite laminates. This paper proposes a novel prediction model 'genetic algorithm optimised multi-layer perceptron neural network' (GA-MLPNN) in which genetic algorithm (GA) is integrated with Multi-Layer Perceptron Neural Network. The performance capability of response surface methodology (RSM) and GA-MLPNN in prediction of thrust force is investigated. RSM is also used to evaluate the influence of process parameters (spindle speed, feed rate, point angle and drill diameter) on thrust force. GA is used to optimize the thrust force and its optimization performance is compared with that of RSM. It is observed that the GA-MLPNN is better predicting tool than the RSM model. The investigation in this paper demonstrates that TiN-SC is the best tool for drilling BD CFRP composite laminates as minimum thrust force is developed during its use. © 2016 Elsevier B.V. All rights reserved.