Conference Papers
Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506
Browse
7 results
Search Results
Item Basics on Categorizing Travel-Time-Based Degrees of Satisfaction Using Triangular Fuzzy-Membership Functions(Elsevier B.V., 2020) Anand, A.; George, V.; Kanthi, R.; Tagore, M.; Padmashree, M.S.The travel desires of trip-makers in urban activity centres depend mainly on the location of residential areas, proximity to various activity centres, household characteristics, and socio-economic factors that influence the choice of travel modes. Decision-making with regard to the choice of a particular mode of travel is fuzzy in nature, and seldom follows a rigid rule-based approach. In this context, the fuzzy-logic approach was considered since it could handle inherent randomness in decision-making related to mode-choice. The present study focuses on the application of this technique making use of revealed preference survey data collected through CES and MVA Systra, later compiled and corrected in various stages at NITK. The difference between the actual travel time by a particular mode, and the theoretical travel time based on average vehicular speeds was used as an important indicator in determining the degrees of satisfaction of the trip-maker. This indicator was computed, and fitted using a normal distribution. It was assumed that indicator values between μ-3σ and μ could be considered for the category of satisfied trip-makers according to the three sigma rule where μ is the mean indicator value, and σ represents the standard deviation. The computed values of the indicators were used in classifying the data into 6 categories of degrees of satisfaction that formed the basic framework for modelling using fuzzy-logic technique. This paper aims at understanding the basic mathematical computations involved in defuzzification using the centroid method for triangular membership functions, and provides a comparison with results obtained using MATLAB. © 2020 The Authors. Published by Elsevier B.V.Item Quantum Machine Learning: A Review and Current Status(Springer Science and Business Media Deutschland GmbH, 2021) Mishra, N.; Kapil, M.; Rakesh, H.; Anand, A.; Mishra, N.; Warke, A.; Sarkar, S.; Dutta, S.; Gupta, S.; Prasad Dash, A.; Gharat, R.; Chatterjee, Y.; Roy, S.; Raj, S.; Kumar Jain, V.; Bagaria, S.; Chaudhary, S.; Singh, V.; Maji, R.; Dalei, P.; Behera, B.K.; Mukhopadhyay, S.; Panigrahi, P.K.Quantum machine learning is at the intersection of two of the most sought after research areas—quantum computing and classical machine learning. Quantum machine learning investigates how results from the quantum world can be used to solve problems from machine learning. The amount of data needed to reliably train a classical computation model is evergrowing and reaching the limits which normal computing devices can handle. In such a scenario, quantum computation can aid in continuing training with huge data. Quantum machine learning looks to devise learning algorithms faster than their classical counterparts. Classical machine learning is about trying to find patterns in data and using those patterns to predict further events. Quantum systems, on the other hand, produce atypical patterns which are not producible by classical systems, thereby postulating that quantum computers may overtake classical computers on machine learning tasks. Here, we review the previous literature on quantum machine learning and provide the current status of it. © 2021, Springer Nature Singapore Pte Ltd.Item CL-NERIL: A Cross-Lingual Model for NER in Indian Languages (Student Abstract)(Association for the Advancement of Artificial Intelligence, 2022) Prabhakar, A.; Majumder, G.S.; Anand, A.Developing Named Entity Recognition (NER) systems for Indian languages has been a long-standing challenge, mainly owing to the requirement of a large amount of annotated clean training instances. This paper proposes an end-to-end framework for NER for Indian languages in a low-resource setting by exploiting parallel corpora of English and Indian languages and an English NER dataset. The proposed framework includes an annotation projection method that combines word alignment score and NER tag prediction confidence score on source language (English) data to generate weakly labeled data in a target Indian language. We employ a variant of the Teacher-Student model and optimize it jointly on the pseudo labels of the Teacher model and predictions on the generated weakly labeled data. We also present manually annotated test sets for three Indian languages: Hindi, Bengali, and Gujarati. We evaluate the performance of the proposed framework on the test sets of the three Indian languages. Empirical results show a minimum 10% performance improvement compared to the zero-shot transfer learning model on all languages. This indicates that weakly labeled data generated using the proposed annotation projection method in target Indian languages can complement well-annotated source language data to enhance performance. Our code is publicly available at https://github.com/aksh555/CL-NERIL. © © 2022, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.Item A Survey on Threat Intelligence Techniques for Constructing, Detecting, and Reacting to Advanced Intrusion Campaigns(Springer, 2023) Anand, A.; Singhal, M.; Guduru, S.; Chandavarkar, B.R.The rise of intrusion has increased the need for cybersecurity in various organizations. A set of these intrusions by an adversary against a particular organization are called intrusion campaigns. Threat intelligence techniques help detect and respond to intrusion attempts and help organizations set up a framework that can secure their services and interests. This chapter surveys different parameters and resources required to construct such a threat intelligence technique for an organization. Furthermore, the chapter discusses the various cases and models of an Intrusion Detection System (IDS) and Intrusion Response System (IRS) along with their comparison using the security resources collected during the construction of a Threat Intelligence model. All of this combined forms the threat intelligence technique. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.Item Trusted Federated Learning Framework for Attack Detection in Edge Industrial Internet of Things(Institute of Electrical and Electronics Engineers Inc., 2023) Singh, M.P.; Anand, A.; Prateek Janaswamy, L.A.; Sundarrajan, S.; Gupta, M.The edge Industrial Internet of Things (IIoT) is highly vulnerable to attacks due to the vast number of connected devices and the lack of security features. Attacks in edge IIoT can lead to significant damage, including data theft, malfunctioning, and privacy breaches. Federated Learning (FL) is a promising approach to detecting attacks by utilizing edge devices’ collective intelligence. FL allows devices to collaboratively learn from multiple devices’ data without centralized sharing, which preserves data privacy and reduces communication costs. However, FL has vulnerabilities that can compromise model accuracy, privacy, and security. Trusted FL is essential for collaboration among multiple edge IIoT devices while preserving data privacy and security. Trust plays a critical role in the success of FL, as edge IIoT devices must trust that the models are accurately learning and that their data is protected. To address this, we propose an FL framework that uses Federated Averaging (FedAvg) and Convolutional Neural Network (CNN) to detect attacks in edge IIoT. We also propose a mechanism to calculate trust for appropriate edge IIoT device selection by measuring each device’s (a.k.a client’s) performance during model training. The proposed edge IIoT device selection method, client selection, can fairly select clients for model training and improve trust in the entire system. Although the proposed FL approach does not outperform the centralized ResNet-18 CNN model on experimental analysis, improving its performance can be a promising solution for detecting attacks in edge IIoT. © 2023 IEEE.Item PUF-Based Ownership Transfer Using Blockchain(Springer Science and Business Media Deutschland GmbH, 2025) Cunha, T.B.D.; Manjappa, M.; Singh, V.; Anand, A.Counterfeiting of electronic components in the branded products is one of the most important and difficult issues to deal with in national/international markets along with the trusted ownership transfer of the product. Today we have to trust an individual while buying a product believing that the product is not tampered. But, we do not have any trusted source which can back this claim. This creates a lot of speculation in the market. For a long time RFID tags were used to find the anti counterfeits in the supply chain, but the problem with the RFID tag is that they can be cloned and hence the authenticity of the tags over the network is questionable. Hence, in order to counter this, we are leveraging blockchain technology to build a novel ownership transfer protocol where the ownership transfer mechanism is secured and authenticated using Physically Unclonable Functions (PUF). The genuinity of the product is checked by PUF by using Challenge Response check during the ownership transfer. Further, the ownership transfer history of the particular product is also maintained in the blockchain which helps the buyer to get more details on the product. The proposed blockchain architecture also provides a temporary ownership transfer option for the owners during servicing or leasing. The proposed architecture is implemented in ethereum blockchain platform and tested for its efficiency. The architecture is found to be promising. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.Item Modal Split and Cost-Sensitivity Analysis for Various Travel Modes Using Calibrated Parameters in NL Modeling(Elsevier B.V., 2025) George, V.; Anand, A.The present study demonstrates the use of the nested-logit (NL) approach in modal-split modeling with input related to the total cost of travel determined based on the cost of travel per km, and the cost of travel time per minute incurred by trip-makers. The calibrated best value of the scaling parameter that simulated the actual travel pattern as in Amritsar city in India was identified based on a trial-and-error approach, followed by cost-based sensitivity analyses. The study revealed that an increase in the combined costs of travel by private modes including intermediate public transport (IPT) modes resulted in a higher estimated ridership by public modes of travel such as mini-buses. Similar analyses were performed to estimate the ridership for private modes Vs IPT modes. One of the key findings of the cost-sensitivity analysis is that even when the total cost of travel by private modes including IPT was increased by up to 35%, the ridership by these modes remained more dominant than that of public modes. In Amritsar, trip-makers rely on independently operated minibuses and IPT alternatives that provide shared rides in place of buses. The insights provided can help formulate policies for promoting public transport modes. © 2024 The Authors.