A5049739048- Speech and Audio Processing
- Music and Audio Processing
- Advanced Vision and Imaging
- Music Technology and Sound Studies
- Advanced Radiotherapy Techniques
- Machine Learning and Algorithms
- Radiation Therapy and Dosimetry
- Advanced Image and Video Retrieval Techniques
- Topic Modeling
- Face recognition and analysis
- Neuroscience and Music Perception
- Neural Networks and Applications
- Medical Imaging Techniques and Applications
- Machine Learning and Data Classification
- Image Retrieval and Classification Techniques
- Face and Expression Recognition
- Natural Language Processing Techniques
- Radiation Effects and Dosimetry
- Advanced Text Analysis Techniques
- Video Analysis and Summarization
- Software System Performance and Reliability
- Cosmology and Gravitation Theories
- Speech and dialogue systems
- Multimodal Machine Learning Applications
- Complex Network Analysis Techniques
Medical College and Hospital, Kolkata
2012-2023
KPC Medical College and Hospital
2023
University of Florida
2016-2021
Microsoft Research (United Kingdom)
2004-2021
Columbus Oncology and Hematology Associates
2019
Seattle University
2018
University of Houston
2016
Microsoft (United States)
2006-2015
Cornell University
2015
Chittaranjan National Cancer Institute
2012
This paper describes a method for the robust tracking of rigid head motion from video. uses 3D ellipsoidal model and interprets optical flow in terms possible motions model. is to large angular translational not subject singularities 2D The has been successfully applied heads with variety shapes, hair styles, etc. also advantage accurately capturing parameters head. accuracy shown through comparison ground truth synthetic sequence (a rendered animation head). In addition, small variations...
We introduce a new approach to the machine-assisted grading of short answer questions. follow past work in automated by first training similarity metric between student responses, but then go on use this group responses into clusters and subclusters. The resulting groupings allow teachers grade multiple with single action, provide rich feedback groups similar answers, discover modalities misunderstanding among students; we refer amplification grader effort as “powergrading.” develop means...
We introduce MySong, a system that automatically chooses chords to accompany vocal melody. A user with no musical experience can create song instrumental accompaniment just by singing into microphone, and experiment different styles chord patterns using interactions designed be intuitive non-musicians.
Igor Labutov, Sumit Basu, Lucy Vanderwende. Proceedings of the 53rd Annual Meeting Association for Computational Linguistics and 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.
An overwhelming number of news articles are available every day via the internet. Unfortunately, it is impossible for us to peruse more than a handful; furthermore difficult ascertain an article's social context, i.e., popular, what sorts people reading it, etc. In this paper, we develop system address problem in restricted domain political by harnessing implicit and explicit contextual information from blogosphere. Specifically, track thousands blogs they cite, collapsing that have highly...
We describe tools that use measurements from video for the extraction of facial modeling and animation parameters, head tracking, real time interactive animation. These share common goals but rely on varying details physical geometric in their input measurement system. Accurate involves fine geometry muscle coarticulation. By coupling pixel by surface motion to a physically based face model control model, we have been able obtain detailed spatio temporal records both displacement each point...
In comparison to multiple choice or other recognition-oriented forms of assessment, short answer questions have been shown offer greater value for both students and teachers; they can improve retention knowledge, while teachers provide more insight into student understanding. Unfortunately, the same open-ended nature which makes them so valuable also difficult grade at scale. To address this, we propose a cluster-based interface that allows read, grade, feedback on large groups answers once....
Researchers have noted conflicting trends in collaboration technologies between delivering more information on larger displays and exploiting mobility smaller devices. Large, shared provide greater choice the presentation of information, but mobile devices offer flexibility access information. We describe a platform that leverages best both worlds by allowing multiple users to interact with large, display using their own personal devices, such as cell phone, laptop, or wireless PDA....
Given a classification task, what is the best way to teach resulting boundary human? While machine learning techniques can provide excellent methods for finding boundary, including selection of examples in an online setting, they tell us little about how we would human same task. We propose investigate problem example and presentation context teaching humans, explore variety mechanisms interests may work best. In particular, begin with baseline random then examine combinations several...
Abstract Chronic hepatitis B virus (HBV) infection is partly responsible for hepatitis, fatty liver disease and hepatocellular carcinoma (HCC). HBV core protein (HBc), encoded by the genome, may play a significant role in life cycle. However, function of HBc occurrence development still unclear. To investigate underlying mechanisms, HBc-transfected HCC cells were characterized multi-omics analyses. Combining proteomics metabolomics analyses, our results showed that promoted expression...
Physiologically based pharmacokinetic (PBPK) models are increasingly used to support pediatric dose selection for small molecule drugs. In contrast, only a few PBPK therapeutic antibodies have been published recently, and the knowledge on maturation of processes relevant antibody pharmacokinetics (PK) is limited compared molecules. The aim this study was, thus, evaluate predictions from children which were scaled adults in order identify respective gaps. For this, we generic model...
A real-time system for tracking and modeling of faces using an analysis-by-synthesis approach is presented. 3D face model texture-mapped with a head-on view the face. Feature points in face-texture are then selected based on image Hessians. The rendered tracked incoming video normalized correlation. result fed into extended Kalman filter to recover camera geometry, head pose, structure from motion. This information used rigidly move render next needed tracking. Every point filter's estimated...
We address the problem of tracking and reconstructing 3D human lip motions from a 2D view. This is challenging due both to complex nature minimal data available raw video stream face. counter these difficulties with statistical approaches. first build physically-based model lips train it cover only subspace motions. then track this in by finding shape within that maximizes posterior probability given observed features. In study, features are likelihoods non-lip color classes: we iteratively...
We present a novel method for simultaneous voicing and speech detection based on linked-HMM architecture, with robust features that are independent of the signal energy. Because this approach models change in dynamics between nonspeech regions, it is to low sampling rates, significant levels additive noise, large distances from microphone. demonstrate performance our variety testing conditions also compare other methods reported literature.
We propose a three-stage pixel based visual front end for automatic speechreading (lipreading) that results in improved recognition performance of spoken words or phonemes. The proposed algorithm is cascade three transforms applied to three-dimensional video region interest contains the speaker's mouth area. first stage typical image compression transform achieves high "energy", reduced-dimensionality representation data. second linear discriminant analysis data projection, which...