DSDilip Sarkar

Dilip Sarkar

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Dilip Sarkar has a total of 31 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Evolution of Artificial Intelligence: From Symbolic to Machine Learning to Neural Learning to Modern AI

    by Dilip Sarkar

    The year 2022 marks the beginning of AI's mass popularity; a report by Menlo Ventures estimated about 500-600 million daily users. Yet it took 70-75 years and four evolutionary cycles for AI to reach this point. The early phase relied on symbolic, rule-based systems, but embedding domain knowledge as logical expressions or IF-THEN rules proved unscalable, as true scalability required extracting knowledge directly from data. Statistical analysis and optimization methods gave rise to machine learning (ML), collectively known as statistical ML, providing faster and more practical ways to build intelligent systems. Still, ML systems fell short of human-like intelligence. In parallel, artificial neurons, inspired by their biological counterparts, were connected into networks and trained on example datasets. Though not mathematically rigorous, these networks began to show traces of human-like intelligence. The next major advance came with deep learning architectures, which enabled embedding knowledge from vast corpora of data into models capable of human-like performance, giving rise to the modern AI we know today.

  • Open Access

    Semantic Uncertainty Quantification of Hallucinations in LLMs: A Quantum Tensor Network Based Method

    by Pragatheeswaran Vipulanandan, Kamal Premaratne, Dilip Sarkar

    Large language models (LLMs) exhibit strong generative capabilities but remain vulnerable to confabulations, fluent yet unreliable outputs that vary arbitrarily even under identical prompts. Leveraging a quantum tensor network based pipeline, we propose a quantum physics inspired uncertainty quantification framework that accounts for aleatoric uncertainty in token sequence probability for semantic equivalence based clustering of LLM generations. This offers a principled and interpretable scheme for hallucination detection. We further introduce an entropy maximization strategy that prioritizes high certainty, semantically coherent outputs and highlights entropy regions where LLM decisions are likely to be unreliable, offering practical guidelines for when human oversight is warranted. We evaluate the robustness of our scheme under different generation lengths and quantization levels, dimensions overlooked in prior studies, demonstrating that our approach remains reliable even in resource constrained deployments. A total of 116 experiments on TriviaQA, NQ, SVAMP, and SQuAD across multiple architectures including Mistral-7B, Mistral-7B-instruct, Falcon-rw-1b, LLaMA-3.2-1b, LLaMA-2-13b-chat, LLaMA-2-7b-chat, LLaMA-2-13b, and LLaMA-2-7b show consistent improvements in AUROC and AURAC over state of the art baselines.

  • Open Access

    A Quantum Tensor Network-Based Viewpoint for Modeling and Analysis of Time Series Data

    by Pragatheeswaran Vipulananthan, Kamal Premaratne, Dilip Sarkar, Manohar N. Murthi

    Accurate uncertainty quantification is a critical challenge in machine learning. While neural networks are highly versatile and capable of learning complex patterns, they often lack interpretability due to their ``black box'' nature. On the other hand, probabilistic ``white box'' models, though interpretable, often suffer from a significant performance gap when compared to neural networks. To address this, we propose a novel quantum physics-based ``white box'' method that offers both accurate uncertainty quantification and enhanced interpretability. By mapping the kernel mean embedding (KME) of a time series data vector to a reproducing kernel Hilbert space (RKHS), we construct a tensor network-inspired 1D spin chain Hamiltonian, with the KME as one of its eigen-functions or eigen-modes. We then solve the associated Schr{ö}dinger equation and apply perturbation theory to quantify uncertainty, thereby improving the interpretability of tasks performed with the quantum tensor network-based model. We demonstrate the effectiveness of this methodology, compared to state-of-the-art ``white box" models, in change point detection and time series clustering, providing insights into the uncertainties associated with decision-making throughout the process.

  • Open Access

    Novel Deep Neural Network Classifier Characterization Metrics with Applications to Dataless Evaluation

    by Nathaniel Dean, Dilip Sarkar

    The mainstream AI community has seen a rise in large-scale open-source classifiers, often pre-trained on vast datasets and tested on standard benchmarks; however, users facing diverse needs and limited, expensive test data may be overwhelmed by available choices. Deep Neural Network (DNN) classifiers undergo training, validation, and testing phases using example dataset, with the testing phase focused on determining the classification accuracy of test examples without delving into the inner working of the classifier. In this work, we evaluate a DNN classifier's training quality without any example dataset. It is assumed that a DNN is a composition of a feature extractor and a classifier which is the penultimate completely connected layer. The quality of a classifier is estimated using its weight vectors. The feature extractor is characterized using two metrics that utilize feature vectors it produces when synthetic data is fed as input. These synthetic input vectors are produced by backpropagating desired outputs of the classifier. Our empirical study of the proposed method for ResNet18, trained with CAFIR10 and CAFIR100 datasets, confirms that data-less evaluation of DNN classifiers is indeed possible.

  • Open Access

    Benefits and Limitations of Web3

    by Collin Connors, Dilip Sarkar

    Web3 provides users and service providers several benefits not found in Web2. However, despite the benefits provided, Web3 faces several obstacles that prevent the paradigm from gaining widespread adoption. Developers should understand the benefits and limitations of the technology in order to create more accessible Web3 smart applications.

  • DNN Feature Map Gram Matrices for Mitigating White-Box Attacks on Image Classification

    by Nathaniel Dean, Dilip Sarkar

    Deep convolutional neural networks accurately classify a diverse range of natural images, but may be easily deceived when designed, imperceptible perturbations are embedded in the images. In this paper, we design a multi-pronged training, input transformation, and image ensemble system that is attack agnostic and not easily estimated. Our system incorporates two novel features. The first is a transformation layer that iteratively filters the input image’s feature maps based on their Gram matrix representational differences to those of randomly sampled training instances from each class, which creates an ensemble of filtered input image copies. The second is a classification system that treats this filtered ensemble as a voting committee that focuses on only those predictions that changed from the original prediction on the unfiltered input image. Our evaluations on the CIFAR10 dataset show our system improves the robustness of an undefended network against a variety of bounded and unbounded white-box attacks under different distance metrics, while sacrificing little accuracy on clean images. Against adaptive full-knowledge attackers creating end-to-end attacks, our system successfully augments the existing robustness of adversarially trained networks, for which our methods are most effectively applied.

  • Open Access

    A Perturbation Resistant Transformation and Classification System for Deep Neural Networks

    by Nathaniel Dean, Dilip Sarkar

    Deep convolutional neural networks accurately classify a diverse range of natural images, but may be easily deceived when designed, imperceptible perturbations are embedded in the images. In this paper, we design a multi-pronged training, input transformation, and image ensemble system that is attack agnostic and not easily estimated. Our system incorporates two novel features. The first is a transformation layer that computes feature level polynomial kernels from class-level training data samples and iteratively updates input image copies at inference time based on their feature kernel differences to create an ensemble of transformed inputs. The second is a classification system that incorporates the prediction of the undefended network with a hard vote on the ensemble of filtered images. Our evaluations on the CIFAR10 dataset show our system improves the robustness of an undefended network against a variety of bounded and unbounded white-box attacks under different distance metrics, while sacrificing little accuracy on clean images. Against adaptive full-knowledge attackers creating end-to-end attacks, our system successfully augments the existing robustness of adversarially trained networks, for which our methods are most effectively applied.

  • Open Access

    Single Sign-On: A Solution Approach to Address Inefficiencies During Sign-Out Process

    by Lokesh Saravanan Ramamoorthi, Dilip Sarkar

    In a Single Sign-on (SSO) environment, an Identity Provider (IDP) authenticates a user for the first Service Provider (SP). The IDP creates an active IDP session and stores its information in the user’s web browser. Each SP also creates and maintains one active service session. Using state-transition diagrams, we illustrate sign-in and sign-out processes. An information security vulnerability situation is created because users are unaware of an active IDP session in the user’s browser and signs-out only from SP sessions. One solution to this problem is educating users. Another solution is to implement the SSO that ensures the termination of the IDP session as soon as user signs-out from all services that the IDP authenticated. The first solution appears to be simple, but practically an impossible task to educate millions of web based SSO users worldwide. The second solution is better because one good implementation solves the problem for all users. In this article, we propose several solutions for terminating the hidden active IDP session. Also, we review the data storage-methods commonly used for storing information of SP and IDP sessions in the browsers. Moreover, we propose a browser extension for conveniently and efficiently managing active SP and IDP sessions. In our proposed browser extension, we have recommended IndexedDB browser storage for storing active session information. We believe our proposed browser extension is simple, but efficient solution for eliminating hidden active IDP session.