Optical control of a metastable phase in the charge density wave Mott insulator 1T-TaS2 investigated using time- and angle-resolved photoemission spectroscopy Tanusree Saha, Arindam Pramanik, Rokaya Osama, Fabio Frassetto, Damjan Krizmancic, Luca Poletto, Arun Ravindran, Barbara Ressel, Primož Rebernik Ribič, Giovanni De Ninno Physical Review B, 2026 Light-induced metastable phases are exotic, long-lived out-of-equilibrium states of matter. Optical control offers a powerful approach to engineering these phases, enabling dynamic tuning of their electronic and structural properties. Using time- and angle-resolved photoemission spectroscopy, we investigate the emergence of a metastable phase induced by a strong infrared pump in the charge-density-wave (CDW)-Mott insulator 1 T − TaS 2 . Furthermore, we demonstrate how its properties can be optically manipulated by varying the photoexcitation strength. A long-lived stabilization of the renormalized electronic band structure serves as a signature of the metastable phase. It displays a relaxed periodic lattice distortion (PLD) and primarily lattice-driven dynamics. The emergence of a new dispersive band in the vicinity of the Hubbard bands reveals the formation of a novel band structure unique to the metastable phase. Our pump-fluence-dependent studies reveal a threshold (incident) fluence F C ∼ 1.3 mJ / cm 2 for inducing the metastable phase, above which the band renormalization continuously evolves with increasing fluence. For F ⩽ 3.4 mJ / cm 2 , stronger photoexcitation progressively drives the phase to higher energies, accompanied by a more relaxed PLD and reduced CDW amplitude. The properties of the metastable phase are strongly influenced by the transient dynamics at each fluence, and the associated fast timescales suggest that the intrinsic CDW amplitude mode remains unaffected by optical manipulation. These findings highlight the potential of optical control in tuning the properties of metastable phases in quantum materials, offering new insights into the manipulation of CDW systems and paving the way for future investigations in nonequilibrium phase engineering.
A Domain-Specific Framework for Priority-Based Modeling and Optimization of Non-Functional Requirements Beyond Binary Constraints Anirban Bhar, Sankhayan Choudhury, Sabnam Sengupta, Suchismita Maiti, Shubhendu Banerjee, Suparna Dasgupta, Soumyabrata Saha, Tanusree Saha, Amrut Ranjan Jena, Soumya Bhattacharyya IEEE Access, 2026 Non-Functional Requirements (NFRs) such as performance, security, reliability, and usability significantly influence the overall quality of software systems. However, traditional NFR engineering approaches commonly treat these requirements as binary constraints, i.e., either fully satisfied or not, which limits systematic trade-off reasoning even in conventional single-system software architectures. This limitation becomes critical in complex, dynamic domains such as IoT and cyber-physical systems, where multiple interdependent NFRs must coexist without degrading system performance. To address this challenge, this paper proposes a domain-specific framework for modeling and optimizing priority-based NFRs beyond binary constraints. The framework integrates standard well-recognized quality models like ISO/IEC 25010 and FURPS+, to identify domain-specific NFRs, which are then modeled as soft goals with graded degrees of satisfaction. Priority weights, utility-based reasoning, and multi-objective optimization are integrated to manage conflicts and trade-offs among competing NFRs while preserving system stability. Unlike traditional approaches that enforce a rigidly defined subset of NFRs, the proposed framework enables the combination of multiple NFRs proportionally according to their priority and contextual relevance. The framework is applicable to diverse software-intensive systems including IoT, cloud platforms, healthcare systems, and cyber-physical environments. This work contributes a systematic, scalable, and adaptive approach to proactive NFR management suitable for modern software systems. The concrete research problem addressed is the failure of binary NFR handling in dynamic adaptation settings, leading to over-constrained designs or degraded quality. The framework guarantees improved global utility (15-25% in simulations) and adaptation success (>90%) while maintaining performance thresholds, with low runtime overhead (~O(n<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) for n NFRs).
CRYPTX Novel Approach for Data Encryption With a Customized Solution for Smarter Human-Computer Interaction Tanusree Saha, Prolay Ghosh, Sumit Das, Nehal Ojha, Kumar Vishal, Suraj Kumar Singh Driving Smarter Human Computer Interaction with Multidisciplinary Personalized Systems, 2025 With the increasing need for secure communication and data protection in the digital age, cryptography and network security protocols have emerged as crucial tools for safeguarding information from unauthorized access. This paper proposes a novel approach that combines cryptography and steganography to create a customized solution for securing data with a smart human-computer interaction system. The paper begins with an in-depth analysis of various cryptographic algorithms, strengths and weaknesses of each algorithms has been evaluated, and a customized encryption scheme is designed that leverages the advantages of different algorithms to enhance security. Next, Steganography is integrated into encryption with a custom algorithm that hides encrypted data in innocent cover objects like images or audio files, tailored to application-specific requirements. This work contributes to the field of information security by introducing a novel approach that offers increased protection against unauthorized access and data breaches with smart human computer interaction facilities
A Heart Disease Prediction Model Applying Deep Learning and Combined Fuzzy-Mutual Information Based Feature Selection Annwesha Banerjee Majumder, Somsubhra Gupta, Sourav Majumder, Dharmpal Singh, Sumit Das, Tanusree Saha Baghdad Science Journal, 2025 Despite remarkable advancements in medical technology, cardiovascular disease persists as a significant contributor to global mortality. This research addresses the imperative need for timely disease identification through the proposition of a heart disease prediction model, utilizing a merged 1D Convolutional Neural Network (CNN). The main aim of the research is to mitigate the inherent drawbacks of single-layer designs, with a particular emphasis on enhancing hierarchical feature extraction, broadening the model’s receptive fields, and facilitating more efficient non-linear transformations. The dataset has been collected from UCI data repository. The study methodology includes a new feature selection strategy that combines Mutual Information and Fuzzy Logic approaches, offering a subtle viewpoint that is not well covered in the literature at the moment. After undergoing extensive training, the output of two 1D CNNs are merged to provide an impressive average accuracy rate of 85%. The integrated 1D CNN model and Explainable AI methodologies yield promising outcomes in heart disease prediction, affirming the model’s potential as a screening instrument for early identification and intervention. By providing a strong framework for proactive cardiovascular health care, the study makes a substantial contribution to the nexus between medical science and machine learning.
An Intelligent Energy Efficient Dynamic Load Balancing Algorithm Using Machine Learning for Cloud Computing Environment Annwesha Banerjee Majumder, Sushma Mallick, Tanusree Saha, Rupashri Barik, Trishita Ghosh, Chowdhury Md Mizan Conference Proceedings of International Conference on Sustainable Technology in Energy and Power Systems Stepcon 2025, 2025 The process of distributing workloads among computing resources in a cloud computing environment and methodically balancing the network traffic that accesses those resources is referred to as cloud load balancing. Load balancing is essential in cloud computing to provide high availability, performance, and scalability of applications by allocating traffic among several servers or resources. In this paper an intelligent dynamic load balancing algorithm has been proposed for cloud environment. The proposed intelligent dynamic load balancing algorithm optimizes server selection by evaluating energy consumption, CPU capacity, and predicted response time. Using Linear Regression, it predicts response time based on CPU Utilization, Memory usage, Storage usage, Network Bandwidth, Service Latency of multi cloud service composition dataset which achieved 132.294501 MAE. Unlike traditional methods, it dynamically evaluates server suitability using real-time metrics such as load, energy use, and urgency. This methodology can help in proper server utilization by equivalently distributing loads in servers.
Encrypted negative password (ENP) authentication system Namrata Barua, Tanusree Saha, Jui Pattanayak, Prolay Ghosh Interdisciplinary Approaches to AI Internet of Everything and Machine Learning, 2024 Password authentication is among the most effective ways to authenticate for security resiliency. The challenge of securely keeping credentials is one of the most important ones. Based on the Negative Database (NDB), a cryptographic hash function, and symmetric encryption, our study suggests the Encrypted Negative Password (ENP) password protection approach. Methods: This paper suggests a password verification structure for protected password storage that may be effectively incorporated into existing authentication systems. A hybrid method is used to encrypt the client's plain password. Additionally, a password authentication system built on the ENP is presented. Because it is the hash value of each user's password, the secret key in the ENP is almost always unique and does not have to be separately produced and maintained. As a result, passwords can be protected using symmetric encryption.
Anxiety, depression, and stress prediction in modern life using machine learning, taxonomy, applications, and challenges Prolay Ghosh, Tanusree Saha, Ritadrik Chowdhury, Siddhartha Barua Interdisciplinary Approaches to AI Internet of Everything and Machine Learning, 2024 The rise of psychological complication and the importance for fruitful medical treatment have opened the area of an research analysis of machine learning that can be practiced in psychological complication. In this paper we have offered a thorough analysis of techniques of machine learning in Anxiety, Depression and Stress prediction. We have also concentrated on the difficulties, restrictions, and potential future prospects for the use of machine learning in this area. We have collected research journals and studies that enlights on different machine learning techniques in prediction of mental health problems. While doing this review, we adhered to the PRISMA process.. We have listed a total of 15 research studies and articles in this review after comprehensive identification and screening procedures. Then, we classify the research studies based on the different mental health complication such as anxiety, stress and depression etc.
A Study of Crime Analysis Using Machine Learning Suparna Dasgupta, Tanusree Saha, Soumyabrata Saha, Risha Roy, Tanusree Das, Arnab Roy, Chandan Nayak Lecture Notes in Networks and Systems, 2024
Speech Recognition and Emotion Detection using Machine Learning approaches on Ravdess Dataset Tanusree Saha, Soumyabrata Saha, Suparna Dasgupta, Moinak Guha, Priti Roy, Madhurima Majumdar, Suvajit Sarkar, Siddhartha Barua International Conference on Big Data Analytics in Bioinformatics Dabcon 2024, 2024 Speech analysis can be employed in real-time to detect emotions using machine learning. Emotions can be identified by analysing features such as pitch, intensity, and spectral patterns. Various techniques like as neural networks or support vector machines, can be utilised to classify these emotions. It has applications in the fields of psychology, customer service, and computer interface. This work specifically examined the correlation between emotions and speaking. The classification model was trained on the RAVDESS dataset using a combination of deep neural network, Support Vector Machine, and Multi-layer perceptron. The paradigm distinguishes between eight distinct emotions: Neutral, Happy, Angry, Sad, Calm, Surprise, Disgust, and Afraid. The paradigm distinguishes between 8 distinct emotions: Neutral, Calm, Happy, Sad, Angry, Afraid, Disgust, and Surprise. This concept has the potential to transform human-computer interaction by enabling machines to comprehend and reciprocate human emotions. Improved data collection and analysis, uncovering hidden insights in healthcare, enhancing customer service, and developing technology that accurately captures the intricacies of human experience.
