Debajyoti Biswas

Dr. Debajyoti Biswas

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 February 2026 in Synthetic Metals
Verified
Journal Article
Closed access
DBDebajyoti Biswas
Debajyoti Biswas
ARAnuj Rajpoot
Anuj Rajpoot
FIFiheon Imroze
Fiheon Imroze
SVSusy Varughese
Susy Varughese
SDSoumya Dutta
Soumya Dutta

14 October 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
DBDebajyoti Biswas
Debajyoti Biswas
VMVarsha Pillai M
Varsha Pillai M
YRYaramaneni Rithwik
Yaramaneni Rithwik
TCTamojita Chatterjee
Tamojita Chatterjee
ABAjay Badrinath
Ajay Badrinath
In this work, we utilize retrieval augmentation and generation (RAG) models for studying and learning from visually rich journal articles focussing on organic memristors and build a 'laboratory advisor' addressing memristors-related queries from users. The two-level ranking system of RAG model, based on MaxSim and Cross-encoder ranks a set of scientific articles fed to the model, extracts the metadata and the article text and images as embeddings based on user-generated query. Once the articles are ranked, the Google Gemini-based 'laboratory advisor' answers user-queries in natural English language from the knowledge acquired from the set of articles and provides scientific recommendations from other available literature from internet. The model can further analyse the articles' data to generate graphs and other diagrams consolidating literature review output and can help in machine learning-based analyses. Using this domain-agnostic model may reduce the time required by junior researchers for literature review and help in advancing material sciences research.

12 February 2025 in OSF Preprints
Verified
Preprint
Open access
DBDebajyoti Biswas
Debajyoti Biswas
SMShikha Marwaha
Shikha Marwaha
SASriya Atmakuru
Sriya Atmakuru
This research work on the well-known WM811K wafer map dataset intends to integrate the insights provided by wafer manufacturing techniques and the computing power of quantum convolutional neural networks (QCNN) approach. The work begins with demonstrating high accuracy (𝑹 𝟐 =99.32%) of the classical convolutional neural network (CCNN) consisting of three consecutive convolution layers on the WM-811K dataset. In later part of the paper, optimized for multi-class classification, the applied QCNN model consisting of four qubits used for convolutional and pooling layers exhibit 99.38% accuracy. The QCNN model establishes supremacy over CCNN model by reducing time requirement to train the model for classifying defects on the wafer images.

11 February 2025 in TechRxiv
Verified
Preprint
Open access
DBDebajyoti Biswas
Debajyoti Biswas
SMShikha Marwaha
Shikha Marwaha
SASriya Atmakuru
Sriya Atmakuru
This research work on the well-known WM811K wafer map dataset intends to integrate the insights provided by wafer manufacturing techniques and the computing power of quantum convolutional neural networks (QCNN) approach. The work begins with demonstrating high accuracy (𝑹 𝟐 =99.32%) of the classical convolutional neural network (CCNN) consisting of three consecutive convolution layers on the WM-811K dataset. In later part of the paper, optimized for multi-class classification, the applied QCNN model consisting of four qubits used for convolutional and pooling layers exhibit 99.38% accuracy. The QCNN model establishes supremacy over CCNN model by reducing time requirement to train the model for classifying defects on the wafer images.

7 April 2026 in Physica Scripta
Verified
Journal Article
Closed access
SMShikha Marwaha
Shikha Marwaha
DBDebajyoti Biswas
Debajyoti Biswas
SASriya Atmakuru
Sriya Atmakuru
Accurate classification of wafer map defects is crucial for enhancing semiconductor yield and diagnosing process irregularities. Convolutional neural networks (CNNs) have demonstrated impressive performance on the WM-811K wafer map dataset; nevertheless, as structures become more complex, their processing requirements increase. The possibility for more compact representations and effective training through quantum parallelism is shown by recent developments in quantum machine learning (QML), namely quantum neural networks and quantum convolutional models. Inspired by these advancements, this work proposes a hybrid framework for wafer defect classification that combines a Quantum Convolutional Neural Network (QCNN) with a classical CNN (CCNN). As far as we are aware, this is the first time a QCNN has been used for semiconductor wafer map analysis. Experimental results reveal that while the QCNN achieves accuracy ( R 2 = 99.38%) comparable to the CCNN ( R 2 = 99.32%), it greatly decreases training time due to the low-overhead Simultaneous Perturbation Stochastic Approximation (SPSA) optimizer and the compactness of quantum circuits. These results demonstrate the potential of quantum-enhanced architectures for semiconductor analytics in the future as quantum hardware develops.