A5054750708- Domain Adaptation and Few-Shot Learning
- Big Data and Business Intelligence
- Topic Modeling
- Multimodal Machine Learning Applications
- Natural Language Processing Techniques
- COVID-19 diagnosis using AI
- Chronic Kidney Disease and Diabetes
- Climate variability and models
- Advanced Text Analysis Techniques
- Computational Physics and Python Applications
- Anomaly Detection Techniques and Applications
- Meteorological Phenomena and Simulations
- Esophageal and GI Pathology
- Advanced Battery Materials and Technologies
- Gastroesophageal reflux and treatments
- COVID-19 Pandemic Impacts
- TGF-β signaling in diseases
- COVID-19 impact on air quality
- COVID-19 epidemiological studies
- Tropical and Extratropical Cyclones Research
- Liver Disease Diagnosis and Treatment
- Adversarial Robustness in Machine Learning
- Advanced Computational Techniques and Applications
- Hand Gesture Recognition Systems
- Wireless Power Transfer Systems
Indian Institute of Science Bangalore
2021-2024
Marathwada Agricultural University
2024
Deenanath Mangeshkar Hospital and Research Center
2024
Centre for Development of Advanced Computing
2024
Leibniz Institute of Polymer Research
2022-2024
Sanjay Gandhi Post Graduate Institute of Medical Sciences
2018-2023
Dr. D.Y. Patil Vidyapeeth, Pune
2022-2023
Dr. D. Y. Patil Medical College, Hospital and Research Centre
2022-2023
D.Y. Patil University
2023
Amity University
2022
Unsupervised domain adaptation (DA) has gained substantial interest in semantic segmentation. However, almost all prior arts assume concurrent access to both labeled source and unlabeled target, making them unsuitable for scenarios demanding source-free adaptation. In this work <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> , we enable DA by partitioning the task into two: a) source-only generalization b) target Towards former, provide...
Abstract Air pollution poses a significant environmental risk to large cities worldwide, including New Delhi, India's capital. Occurrence of frequent episodes elevated levels air during October March in Delhi and National Capital Tertiary (Delhi-NCT) chokes its ∼32 million residents every year. Current quality models lack the ability accurately predict severe events Delhi-NCT, rendering decision-makers helpless their efforts safeguard public health. To address this, new initiative introduced...
Mobile robots are widely used in the surveillance industry, for military and industrial applications. To carry out tasks like urban search rescue operation, ability to traverse stairs is of immense significance. This paper presents a deep learning based approach stair detection, statistical filtering on images estimation alignment, novel mechanical design an autonomous climbing robot. The primary objective solve problem indoor locomotion over staircases with proposed implementation....
Global water scarcity is a threat that can be alleviated through membrane filtration technologies. However, the widespread adoption of membranes faces significant challenges, primarily due to biofouling. This reason why modifications have been under increasing investigation address fouling issues. Antibacterial membranes, designed combat biofouling by eliminating microorganisms, offer promising solution. Within this study, flat sheet ultrafiltration (UF) with integrated photocatalytic zinc...
Conventional domain adaptation (DA) techniques aim to improve transferability by learning domain-invariant representations; while concurrently preserving the task-discriminability knowledge gathered from labeled source data. However, requirement of simultaneous access and unlabeled target renders them unsuitable for challenging source-free DA setting. The trivial solution realizing an effective original generic mapping improves but degrades task discriminability. Upon analyzing hurdles both...
Conventional Domain Adaptation (DA) methods aim to learn domain-invariant feature representations improve the target adaptation performance. However, we motivate that domain-specificity is equally important since in-domain trained models hold crucial domain-specific properties are beneficial for adaptation. Hence, propose build a framework supports disentanglement and learning of factors task-specific in unified model. Motivated by success vision transformers several multi-modal problems,...
Abstract The advent of weather and climate models has equipped us to forecast or project monsoon rainfall patterns over various spatiotemporal scales; however, utilizing a single model is not usually sufficient yield accurate projection due the inherent uncertainties associated with individual models. An ensemble runs often used for better projections as multimodel (MME). This study analyzes accuracy MME in simulating Indian summer (ISMR) variability using Coupled Model Intercomparison...
We report effect of influence Cd2+ ions on Co-Ni ferrites synthesized by the solid state reaction method. The X-ray diffraction analysis confirmed cubic spinel phase with crystallite size varies between 20-24 nm. SEM images show tetrahedral, octahedral, granular, long bulgy structures variety sizes. Raman spectra for samples x = 0.1 and 0.4 shown peaks corresponding to Alg, Eg T2g which very closely match NiFe2O4 slight variation in peak position as an different chemical formula. Magnetic...
About 15% patients with acute severe ulcerative colitis (UC) fail to respond medical treatment and may require colectomy. An early prediction of response help the treating team their family prepare for alternative options.Data 263 (mean age 37.0 ± 14.0-years, 176, 77% male) UC admitted during a 12-year period were used study predictors using univariate analysis, multivariate linear principal component analysis (PCA), nonlinear artificial neural network (ANN).Of patients, 231 (87.8%)...
Universal Domain Adaptation (UniDA) deals with the problem of knowledge transfer between two datasets domain-shift as well category-shift. The goal is to categorize unlabeled target samples, either into one "known" categories or a single "unknown" category. A major in UniDA negative transfer, i.e. misalignment and classes. To this end, we first uncover an intriguing tradeoff negative-transfer-risk domain-invariance exhibited at different layers deep network. It turns out can strike balance...
The most efficient way of establishing communication through radio waves are no longer suitable in Non-Conventional (No-Co) media viz. underwater & underground due to dynamic channel conditions. To overcome the associated problem, researches have proven Magnetic Induction (MI) method be more these media. As, principle operation this is at low band frequency, thus providing constant conditions for magnetic fields MI a very promising emerging technology, but it suffers from major drawback...
Conventional domain adaptation algorithms aim to achieve better generalization by aligning only the task-discriminative causal factors between a source and target domain. However, we find that retaining spurious correlation non-causal plays vital role in bridging gap improving adaptation. Therefore, propose build framework disentangles supports factor alignment first. We also investigate strong shape bias of vision transformers, coupled with its multi-head attentions, make it suitable...