322 research outputs found
Rare decays of Bc mesons
No full textIn present work, we study the rare semileptonic decays of Bc mesons in the framework of covariant confined quark model. Necessary transition form factors are evaluated in the entire dynamical range of momentum transfer. We further compare our results with the different theoretical predictions
The story of the Soni Ventorum Wind Quintet
No full textThesis (D. Mus. Arts)--University of Washington, 2000The Soni Ventorum Wind Quintet has been the wind quintet-in-residence at the University of Washington School of Music since 1968. Officially founded in 1962, when its members were on the faculty of the Conservatory of Music of Puerto Rico, the group has had a long and stable history. Through their concerts, tours, and recordings, the Soni Ventorum Wind Quintet has established an international reputation. Over the years, many distinguished composers have written works especially for the Soni Ventorum, thus expanding the repertoire of the wind quintet.This study traces the history of the Soni Ventorum Wind Quintet mainly through interviews with the quintet members themselves. This history includes antecedent quintets in which members of the Soni Ventorum Wind Quintet participated (namely, a student quintet at the Curtis Institute, The American Wind Ensemble of Vienna, and the U.S. Seventh Army Symphony Wind Quintet). It covers the founding of the Soni Ventorum Wind Quintet in 1962 at the Conservatory of Music in Puerto Rico through their tenure from 1968 through the present as the wind quintet-in-residence at the University of Washington in Seattle. It gives an account of the establishment of the Soni Ventorum's recording career, their approach to sound and ensemble, their many tours, participation in festivals and competitions, and personnel. The study details the Soni Ventorum's collaborations with colleagues at the University of Washington School of Music, especially the many composers who wrote pieces for the group. One chapter covers ensemble pieces that have been written for the members of the Son! Ventorum Wind Quintet, while another presents wind quintet and quartet arrangements that were prepared by the quintet members themselves. The final chapter provides biographies of the members of the Soni Ventorum Wind Quintet.The Introduction to the study is a brief history of wind quintets. The study concludes with detailed appendices cataloguing the Soni Ventorum Wind Quintet's repertoire, concerts, residencies, tours and a complete discography.At the time of this writing, the author is aware of no other work detailing the history of an established wind quintet
Structured and Sparse Signal Estimation - Fundamental Limits and Error Bounds
No full textUniversity of Minnesota Ph.D. dissertation. May 2015. Major: Electrical Engineering. Advisor: Jarvis Haupt. 1 computer file (PDF); x, 136 pages.Over the past decade, sparsity has become one of the most prevalent themes in signal processing and Big-Data applications. In general, sparsity describes the phenomenon where high-dimensional data can be explained by only a few variables, values, or coefficients. The presence of sparsity often enables efficient algorithms for extracting relevant information from the data. This effort focuses on the theoretical treatment of specialized sensing and inference techniques that exploit sparsity and other forms of structured low-dimensional representations. The first part of this work focuses on noisy matrix estimation and completion problems. We consider the problem of estimating matrices that adhere to a "sparse-factor model" decomposition -- matrices that may be accurately described by a product of two matrices, one of which is sparse -- from noisy observations, where the noise is modeled as random and may arise from any of a number of various likelihood models (e.g., Gaussian, Poisson, Laplace, and even one-bit models). Sparse-factor models can be used to describe collections of vectors that reside in a union of linear subspaces, and can be viewed as a powerful generalization the widely-used principal component analysis technique, which assumes data reside on or near a single subspace. We establish estimation error guarantees for sparse-factor matrix estimation problems (where a noisy observation of each matrix entry is observed) and matrix completion problems (where only a subset of elements is observed, each corrupted by noise), and describe an efficient algorithm for performing inference in problems of this form. HASH(0x2f1988c) In the second part of this work, we examine and quantify the benefits of "adaptive sensing" techniques, which employ data-dependent feedback in the data acquisition process, in the context of a structured sparse inference task. This work is motivated by a desire to formally exploit the structural characteristics and dependencies present in the wavelet representations of many natural images. We devise an efficient and provably optimal (in a minimax sense) adaptive acquisition method for estimating the locations of the significant wavelet coefficients from noisy observations. Our results demonstrate the significant improvements that can be obtained when leveraging the inherent structural dependencies in the sparse representation of the signal to be acquired and incorporating feedback in the measurement process, relative to the best possible methods that utilize either structural information or adaptivity alone. Overall, our results provide essential new insights into the virtues of adaptive data acquisition in sparse inference tasks.Soni, Akshay. (2015). Structured and Sparse Signal Estimation - Fundamental Limits and Error Bounds. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/175341
PULSE-SMART: Pulse-Based Arrhythmia Discrimination Using a Novel Smartphone Application
No full textCo-author Apurv Soni is a medical student in the MD/PhD Program at UMass Medical School.BACKGROUND: Atrial fibrillation (AF) is a common and dangerous rhythm abnormality. Smartphones are increasingly used for mobile health applications by older patients at risk for AF and may be useful for AF screening. OBJECTIVES: To test whether an enhanced smartphone app for AF detection can discriminate between sinus rhythm (SR), AF, premature atrial contractions (PACs), and premature ventricular contractions (PVCs). METHODS: We analyzed two hundred and nineteen 2-minute pulse recordings from 121 participants with AF (n = 98), PACs (n = 15), or PVCs (n = 15) using an iPhone 4S. We obtained pulsatile time series recordings in 91 participants after successful cardioversion to sinus rhythm from preexisting AF. The PULSE-SMART app conducted pulse analysis using 3 methods (Root Mean Square of Successive RR Differences; Shannon Entropy; Poincare plot). We examined the sensitivity, specificity, and predictive accuracy of the app for AF, PAC, and PVC discrimination from sinus rhythm using the 12-lead EKG or 3-lead telemetry as the gold standard. We also administered a brief usability questionnaire to a subgroup (n = 65) of app users. RESULTS: The smartphone-based app demonstrated excellent sensitivity (0.970), specificity (0.935), and accuracy (0.951) for real-time identification of an irregular pulse during AF. The app also showed good accuracy for PAC (0.955) and PVC discrimination (0.960). The vast majority of surveyed app users (83%) reported that it was "useful" and "not complex" to use. CONCLUSION: A smartphone app can accurately discriminate pulse recordings during AF from sinus rhythm, PACs, and PVCs.MD/Ph
Association of common mental disorder symptoms with health and healthcare factors among women in rural western India: results of a cross-sectional survey
No full textFirst author Apurv Soni is a medical student in the MD/PhD Program at UMass Medical School.OBJECTIVES: Information about common mental disorders (CMD) is needed to guide policy and clinical interventions in low-income and middle-income countries. This study's purpose was to characterise the association of CMD symptoms with 3 inter-related health and healthcare factors among women from rural western India based on a representative, cross-sectional survey. SETTING: Surveys were conducted in the waiting area of various outpatient clinics at a tertiary care hospital and in 16 rural villages in the Anand district of Gujarat, India. PARTICIPANTS: 700 Gujarati-speaking women between the ages of 18-45 years who resided in the Anand district of Gujarat, India, were recruited in a quasi-randomised manner. PRIMARY AND SECONDARY OUTCOMES MEASURES: CMD symptoms, ascertained using WHO's Self-Reporting Questionnaire-20 (SRQ-20), were associated with self-reported (1) number of healthcare visits in the prior year; (2) health status and (3) portion of yearly income expended on healthcare. RESULTS: Data from 658 participants were used in this analysis; 19 surveys were excluded due to incompleteness, 18 surveys were excluded because the participants were visiting hospitalised patients and 5 surveys were classified as outliers. Overall, 155 (22·8%) participants screened positive for CMD symptoms (SRQ-20 score ≥8) with most (81.9%) not previously diagnosed despite contact with healthcare provider in the prior year. On adjusted analyses, screening positive for CMD symptoms was associated with worse category in self-reported health status (cumulative OR=9.39; 95% CI 5·97 to 14·76), higher portion of household income expended on healthcare (cumulative OR=2·31; 95% CL 1·52 to 3.52) and increased healthcare visits in the prior year (incidence rate ratio=1·24; 95% CI 1·07 to 1·44). CONCLUSIONS: The high prevalence of potential CMD among women in rural India that is unrecognised and associated with adverse health and financial indicators highlights the individual and public health burden of CMD.MD/Ph
Comparing Pre-trained Human Language Models: Is it Better with Human Context as Groups, Individual Traits, or Both?
No full textPre-trained language models consider the context of neighboring words and
documents but lack any author context of the human generating the text.
However, language depends on the author's states, traits, social, situational,
and environmental attributes, collectively referred to as human context (Soni
et al., 2024). Human-centered natural language processing requires
incorporating human context into language models. Currently, two methods exist:
pre-training with 1) group-wise attributes (e.g., over-45-year-olds) or 2)
individual traits. Group attributes are simple but coarse -- not all
45-year-olds write the same way -- while individual traits allow for more
personalized representations, but require more complex modeling and data. It is
unclear which approach benefits what tasks. We compare pre-training models with
human context via 1) group attributes, 2) individual users, and 3) a combined
approach on five user- and document-level tasks. Our results show that there is
no best approach, but that human-centered language modeling holds avenues for
different methods
Estimation error guarantees for Poisson denoising with sparse and structured dictionary models
No full textLevel set estimation from compressive measurements using box constrained total variation regularization
No full textScaling Up and Exploring Marsh (SUEM)
No full textAn ecosystem is a community of plants and animals. They support agricultural needs and provide clean air and water. However, recent environmental changes such as global warming and rising sea level caused a need to study these ecosystems. Ecological problems have the specificity of being dependent on biological elements with complex interactions. New tools are therefore needed to go past biological constraints and take into account the complexity of living ecosystems.
Data acquisition was done for six years at Sapelo Island located in Georgia. The sampling area was approximately 3,200 square meters. Annually around 10,000 images were captured of a sampling area.
Manually examining of a substantial amount of spatially explicit data is very laborious and resource consuming. The idea is to involve the general public in ecology projects, first train them in identifying species and then allowing them to identify these species. Thus, there was a need to develop a systematic way of accumulating, parameterizing and validating the volunteer’s contribution to these projects.
Two ecology projects were developed as Web-based games. The first game, Scaling Up Marsh Science was developed to create a detailed mosaic of the sampling area by stitching multiple overlapping images. Applying automatic methods for image alignment and stitching did not produce precise results due to various factors such as perspective error, moving animals, amount of overlapping, shadows. The second game, Marsh Explorer was developed to get an exact description of the sampling area such as identification of the plants and animals species together with their locations. The final output of this game delivers a detailed summary of the sampling area.
This thesis adds to the previous versions of these games basic gaming features such as player ranking, game levels, awards, guided training, player feedback system, administrative tools, image uploader, and database redesign. To date, approximately 400 volunteer players are contributing to these two projects.Computer Science, Department o
- …
