View: session overviewtalk overview
09:30 | Data Mining - Intellectual & Psychological Integrity |
10:30 | Publishing Ethics and Author Services |
11:00 | Towards Efficient Edge Computing Through Adoption of Reinforcement Learning Strategies: A Review ABSTRACT. This paper contributes towards the mapping of the variants of Reinforcement Learning (RL) techniques to solve the key challenges of Edge Computing (EC) through broadly addressing task handling and Quality of Service (QoS) parameters. EC has bolstered ever since the advent of Industry 4.0 with computationally reliable heterogeneous mobile secured dynamic edge devices powered by an array of multifarious sensors designed on multi-edge hierarchical architectures found a strong footing on the backbone of ably equipped communication protocols to manifest their growth powered by the advent of 5G technology. However, with millions of such edge devices finding its way in a plethora of EC applications, with each having its own set of domain specific challenges, devising a suitable agent so as to sense the environment and learn from it has driven RL find its way as one of the significant tools to make the EC framework intelligent. Here we lay a good understanding of how RL has achieved noteworthy success to solve some of the pressing EC challenges. |
11:15 | Machine Learning based Earthquake Early Warning (EEW) System: A case study of Himalayan Region PRESENTER: Samik Basu ABSTRACT. Seismic sensing and generation of earthquake alarm is an important application for society at large. In this paper, we propose the strategy of extracting earthquake event features parameters τc and Pd from fast-arriving P-wave signals. The said features are used to explore the performances of some of the popular machine learning (ML) based classifiers to compare their performances in triggering an alarm for the Earthquake Early Warning (EEW) system. We explored four different ML classifiers namely Support Vector Machine (SVM), Naive Bayes, K-Nearest Neighbors (KNN), and Logistic Regression so that the best can be applied for the EEW alarm generation. We have used publicly available data from the PESMOS platform of IIT-Roorkee in this work. |
11:30 | Topic Modelling Based Semantic Search PRESENTER: Mrityunjoy Panday ABSTRACT. In today's world, information is needed to be retrieved from plethora of publications and papers for scientific paper writing, paper reviewing. Retrieval of relevant data has become a very tedious job. In this paper we are providing a holistic approach to retrieve information or abstracts from a corpus of published papers and artifacts, by improving the conventional methods of searching abstracts. A reviewer or a scientist can provide phrases, sentences and even a complete abstract to search for the most relevant abstracts in any huge database of publications. To accomplish this, we have implemented topic modelling based semantic search. Thereafter various AI algorithms are used to find the closest and nearest abstracts to any given search abstract. For this paper we have extensively used the PUBMED data. Our solution also tries to provide other inherent features and relations between words and keywords with interesting visualization to help the user refine the search more. |
Dr. Jayanta Yumnam (National Institute of Electronics and Info Technology (NIELIT), Guwahati,GOI, India)
11:00 | PRESENTER: Dileep Kumar Kadali ABSTRACT. Machine learning is the kind of process to turns out to be an undividable fraction of any methodical work is mostly because of the simplicity of use and the simplicity of generation of data. The data generation task is pricey or tedious, the data is usually generated in parallel amid various research groups and they are communal by scientists. For this activity is the initial task, is frequently to cluster the data into several categories to set up which data from what source are related to each other. In this paper, optimization is usually is implemented in use for such a clustering activity is proposed. When the data is accomplished then they are entire jumbled jointly. One way to prepare an optimization difficulty is to primary decide on a number of clusters that the data may be separated to. After that, for each cluster in not many parameters are used variables for the optimization task. The parameters have to explain a similarity role for a cluster. The activity can be achieved through an optimization solve the prediction activity can be achieved using an optimization process. This process is used a Semi-Supervised learning approach and Data Mining like the K-Nearest Neighboring process and form the path route cluster. The prediction parameter of SRGM is approximated based upon these data clusters using Least Square Estimation. |
11:15 | Machine Learning based Automated Process for Predicting the Anomaly in AIS Data PRESENTER: Bharath Ramkrishna ABSTRACT. In this paper, we present an automated process for detecting the anomaly in Automatic Identification System (AIS) data. Machine learning approaches have been employed to automatically detect anomalies in the AIS data. The opensource AIS data is been used to evaluate the performance of the proposed approach. Supervised machine learning approaches like K Nearest Neighbour, Random Forest, Support Vector Machine classifier is employed to predict the anomalies in the AIS data. The AIS data does not contain the ground truth labels and supervised learning algorithms need labeling data, to address this issue, we employed an unsupervised approach to label the data based on the prior information and characteristics of the AIS data. The labeled data is then used to train the supervised machine learning models. The proposed approach with support vector machine classifier has classified the AIS data into normal and anomaly with an accuracy of 96.5% |
11:30 | A Hybrid Machine Learning Model for Estimation of Obesity Levels ABSTRACT. Obesity has always been a problem which has plagued humans for many generations, which,since the 1975, almost doubled to turn into a global epidemic. The current human dependence on technology has contributed to the problem even more, with the effects visibly pronounced in late teenagers and early adults. Researchers till date, have tried numerous ways to determine the factors that cause obesity in early adults. On that frontier, our hybrid machine-learning model uses the help of some supervised and unsupervised data mining methods like Extremely Randomized Trees, Multilayer Perceptron and XGBoost using Python to detect and predict obesity levels and help healthcare professionals to combat this phenomenon. Our dataset is a publicly available dataset in the UCI MachineLearning Repository, containing the data for the estimation of obesity levels in individuals from the countries of Mexico, Peru, and Colombia, based on their eating habits and physical condition. The proposed model heavily utilizes feature engineering methods and introduces the concept of a hybrid model. This work has shown improved results over prior works and extensive studies have been undertaken to preserve the robustness of this model. |
Alok Ranjan Prusty (Ministry of Skill Development and Entrepreneurship, India)
11:00 | Generation of Indian Sign Language Animation from Audio and Video Content using Natural Language Processing PRESENTER: Ankita Patil ABSTRACT. The hearing impaired population in India forms around 1-2 percent of the Indian population. The standard medium of communication among this community is the Indian Sign Language (ISL). For a person who wants to communicate with the hearing impaired individual, it becomes challenging since general people don't really know ISL. Also, for a hearing impaired individual it is difficult for one to understand the content of normal audio or video by ownself as the communication is carried out using regular language in most of the audios and videos. To overcome these difficulties, in this paper we have discussed the design and implementation of an application - Sign Savvy, a system for Audio and Video conversion to ISL. This application allows the user to provide inputs in distinct media forms such as text (English/Hindi), audio and video. The application then converts the corresponding inputs into ISL format and generates animation for the Indian Sign Language. The application also has a provision of recording of the ISL animations generated that can be downloaded if needed by the user. Further in the paper, the system performance results are discussed and conclusion is presented |
11:15 | PRESENTER: Bhavani Srirangam ABSTRACT. Abstract: One of the most popular unsupervised clustering algorithms is the K-Means clustering algorithm which can be used for segmentation to analyse the data. It is a centroid-based algorithm, where it calculates the distances to assign a point to a cluster. Each cluster is associated with a centroid. The selection of initial centroids and the number of clusters play a major role to decide the performance of the algorithm. In this context, many researchers worked on, but they may not reach a goal to cluster the images in minimum runtime. Existing histogram based initial centroid selection methods are used on grayscale images only. Two methods, i.e., Histogram based initial centroids selection and Equalized Histogram based initial centroids selection to cluster colour images have been proposed in this paper. The colour image has been divided into R, G, B, three channels and calculated histogram to select initial centroids for clustering algorithm. This method has been validated on three benchmark images and compared to the existing K-Means algorithm and K-Means++ algorithms. The proposed methods give an efficient result compared to the existing algorithms in terms of run time. |
11:30 | Siamese Network-based system for criminal identification PRESENTER: Soham Dave ABSTRACT. To discourage criminal activities, the large number of CCTV installations throughout the country play a crucial role. Through this paper, we propose an AI-based solution that can leverage these devices to remotely identify and report absconding criminals. Using the one-shot learning approach, we present a face recognition algorithm that yields accurate results even with low training data. The Siamese Network architecture is used to verify if the face embeddings of the image detected is the same as that of the criminal. Two parallel neural networks are designed to take one input each- one being the detected face and the other being an embedding from the dataset. The outputs of the two networks are compared to predict whether the detected face is the same as the input face or not. This algorithm is further integrated with an automated model for updating the information of the recognized criminal into the database along with updating the appropriate law enforcement authorities about the last known whereabouts. |
11:00 | Jal-Kachara Bot Model: A remedy for floating waste on water bodies. PRESENTER: Omkesh Munde ABSTRACT. In India due to the mismanagement of waste, rivers and lakes are getting contaminated. Even "gutters" of the city are channelized into rivers at the end. This polluted water is causing lot of diseases which results in disturbing the aquatic atmosphere. The polluted water becomes smelly and makes hard for people who are living in nearby areas. This smelly water also stops waste collectors to not go into the river for cleaning. The floating waste on water bodies blocks the sun light reaching to aquatic life. Solid waste consisting of plastic carry bags, styrofoam cups, thermocol plates etc find their way into the mangroves, hindering the natural process of water bodies and dirtying the city’s beaches. The proposed system can detect garbage floating on water with the help of camera present on system and the robotic arm attached to the system collects and stores waste in on-board trash bin. The system is easily customizable we can customize its size and arms based on size of river. It can also be control manually with an android app having a user-friendly interface to interact with the system. |
11:15 | A Study Of Parasitic Computing PRESENTER: Rujuta Chopade ABSTRACT. “PARASITE” as the word suggests is an entity that resides on another entity exploiting the resources of the latter. The term “PARASITIC COMPUTING” refers to the technique of using the resources of one computer by another computer without the knowledge of the former. Distributed computing networks turn home users’ computers into part of a virtual supercomputer that can perform time-intensive operations. This Research paper provides an insight into the details of how parasitic computing uses the computation power of the computers connected to the internet in solving complex mathematical problems. This technique was developed by the scientist at the Notre Dame University, Indiana (USA). According to the scientists, the transmission control protocol (TCP), could be used to solve a piece of a mathematical problem whose answer could then be relayed back to the original user. The implementation is discussed with the NP-Complete problem as example. Unlike hackers who exploit flaws to gain direct access to machines, the Notre Dame computer scientists created a virtual computer by using the fundamental components of distributed computing. |
12:40 | Big Data Applications in Power Systems |
14:30 | Holographic Data Representations fot Progressive Retrieval |