A5083194525- Neural dynamics and brain function
- Gene expression and cancer classification
- Functional Brain Connectivity Studies
- Remote Sensing in Agriculture
- Advanced Vision and Imaging
- Remote Sensing and LiDAR Applications
- Chronic Myeloid Leukemia Treatments
- Single-cell and spatial transcriptomics
- Statistical Methods and Inference
- Myeloproliferative Neoplasms: Diagnosis and Treatment
- Hepatitis B Virus Studies
- Chronic Lymphocytic Leukemia Research
- Gene Regulatory Network Analysis
- Bioinformatics and Genomic Networks
- Data Visualization and Analytics
- Bayesian Modeling and Causal Inference
- Morphological variations and asymmetry
- Cancer-related molecular mechanisms research
- Pediatric Hepatobiliary Diseases and Treatments
- Acute Lymphoblastic Leukemia research
- Galaxies: Formation, Evolution, Phenomena
- Digital Media Forensic Detection
- Data Analysis with R
- stochastic dynamics and bifurcation
- Viral gastroenteritis research and epidemiology
University of Notre Dame
2020-2024
Rice University
2018-2020
Carnegie Mellon University
2014-2018
Eye and Ear Foundation
2018
University of Modena and Reggio Emilia
2001
University of Catania
1998
Populations of cortical neurons exhibit shared fluctuations in spiking activity over time. When measured for a pair multiple repetitions an identical stimulus, this phenomenon emerges as correlated trial-to-trial response variability via spike count correlation (SCC). However, counts can be viewed noisy versions firing rates, which vary from trial to trial. From perspective, the SCC becomes version corresponding rate (FRC). Furthermore, magnitude is generally smaller than that FRC and likely...
A major challenge in contemporary neuroscience is to analyze data from large numbers of neurons recorded simultaneously across many experimental replications (trials), where the are counts neural firing events, and one basic problems characterize dependence structure among such multivariate counts. Methods estimating high-dimensional covariation based on $\ell_{1}$-regularization most appropriate when there a small number relatively partial correlations, but often correlations. Furthermore,...
The characterization of fine temporal-resolution land surface dynamics from broadband optical satellite sensors is constrained by sparse acquisitions high-quality imagery; interscene variation in radiometric, phenological, atmospheric, and illumination conditions; subpixel variability heterogeneous environments. In this letter, we address these concerns developing testing the automatic adaptive signature generalization regression (AASGr) algorithm. Provided a robust reference map...
Recent advances in single-cell RNA sequencing (scRNA-seq) technologies have yielded a powerful tool to measure gene expression of individual cells. One major challenge the scRNA-seq data is that it usually contains large amount zero values, which often impairs effectiveness downstream analyses. Numerous imputation methods been proposed deal with these "dropout" events, but this difficult task for such high-dimensional and sparse data. Furthermore, there debates on nature sparsity, about...
Graphical model selection is a seemingly impossible task when many pairs of variables are never jointly observed; this requires inference conditional dependencies with no observations corresponding marginal dependencies. This under-explored statistical problem arises in neuroimaging, for example, different partially overlapping subsets neurons recorded non-simultaneous sessions. We call challenge the "Graph Quilting" problem. study context sparse inverse covariance learning, and focus on...
Factor models are widely used in the analysis of high-dimensional data several fields research. Estimating a factor model, particular its covariance matrix, from partially observed vectors is very challenging. In this work, we show that when structurally incomplete, model likelihood function can be decomposed into product functions multiple partial relative to different subsets data. If these linked together by common parameters, then obtain complete maximum estimates parameters and thereby...
Covariance matrix estimation is an important task in the analysis of multivariate data disparate scientific fields, including neuroscience, genomics, and astronomy. However, modern are often incomplete due to factors beyond control researchers, missingness may prohibit use traditional covariance methods. Some existing methods address this problem by completing matrix, or filling missing entries sample assuming a low-rank structure. We propose novel approach that exploits auxiliary variables...
Researchers continue exploring neurons' intricate patterns of activity in the cerebral visual cortex response to stimuli. The way neurons communicate and optimize their interactions with each other under different experimental conditions remains a topic active investigation. Probabilistic Graphical Models are invaluable tools neuroscience research, as they let us identify functional connections, or conditional statistical dependencies, between neurons. models represent these connections...
Single cell RNA sequencing is a powerful technique that measures the gene expression of individual cells in high throughput fashion. However, due to inefficiency, data unreliable dropout events, or technical artifacts where genes erroneously appear have zero expression. Many imputation methods been proposed alleviate this issue. Yet, effective can be difficult and biased because sparse high-dimensional, resulting major distortions downstream analyses. In paper, we propose completely novel...
Abstract The incredible variety of galaxy shapes cannot be summarized by human defined discrete classes without causing a possibly large loss information. Dictionary learning and sparse coding allow us to reduce the high dimensional space into manageable low continuous vector space. Statistical inference can done in reduced via probability distribution estimation manifold estimation.
The incredible variety of galaxy shapes cannot be summarized by human defined discrete classes without causing a possibly large loss information. Dictionary learning and sparse coding allow us to reduce the high dimensional space into manageable low continuous vector space. Statistical inference can done in reduced via probability distribution estimation manifold estimation.