A5103088414- DNA and Biological Computing
- Tissue Engineering and Regenerative Medicine
- Congenital Diaphragmatic Hernia Studies
- Fibroblast Growth Factor Research
- Advanced Data Storage Technologies
- Algorithms and Data Compression
- Domain Adaptation and Few-Shot Learning
- Machine Learning and Data Classification
- Cellular Automata and Applications
- Cancer Cells and Metastasis
- Error Correcting Code Techniques
- Digestive system and related health
- Topic Modeling
- Advanced biosensing and bioanalysis techniques
- Neonatal Respiratory Health Research
- Natural Language Processing Techniques
- Congenital gastrointestinal and neural anomalies
- Machine Learning and Algorithms
- Clinical Nutrition and Gastroenterology
- 3D Printing in Biomedical Research
- Wnt/β-catenin signaling in development and cancer
- Coding theory and cryptography
- Cancer Research and Treatments
- Congenital heart defects research
- Advanced Memory and Neural Computing
University of Wisconsin–Madison
2021-2023
Stanford University
2017-2021
Stanford Medicine
2020
University of California, Los Angeles
2012-2018
UCLA Health
2014-2017
Children's Hospital of Los Angeles
2005-2015
University of Southern California
2006-2015
California Institute for Regenerative Medicine
2010
Institut Curie
2006
Centre National de la Recherche Scientifique
2006
Knowledge graph (KG) embeddings learn low- dimensional representations of entities and relations to predict missing facts. KGs often exhibit hierarchical logical patterns which must be preserved in the embedding space. For data, hyperbolic methods have shown promise for high-fidelity parsimonious representations. However, existing do not account rich KGs. In this work, we introduce a class KG models that simultaneously capture patterns. Our approach combines reflections rotations with...
Hyperbolic embeddings offer excellent quality with few dimensions when embedding hierarchical data structures like synonym or type hierarchies. Given a tree, we give combinatorial construction that embeds the tree in hyperbolic space arbitrarily low distortion without using optimization. On WordNet, our obtains mean-average-precision of 0.989 only two dimensions, while Nickel et al.'s recent 0.87 200 dimensions. We provide upper and lower bounds allow us to characterize...
As machine learning models continue to increase in complexity, collecting large hand-labeled training sets has become one of the biggest roadblocks practice. Instead, weaker forms supervision that provide noisier but cheaper labels are often used. However, these weak sources have diverse and unknown accuracies, may output correlated labels, label different tasks or apply at levels granularity. We propose a framework for integrating modeling such by viewing them as labeling related sub-tasks...
Short bowel syndrome (SBS) is a devastating condition in which insufficient small intestinal surface area results malnutrition and dependence on intravenous parenteral nutrition. There an increasing incidence of SBS, particularly premature babies newborns with congenital anomalies. Tissue-engineered intestine (TESI) offers therapeutic alternative to the current standard treatment, transplantation, has potential solve its biggest challenges, namely donor shortage life-long immunosuppression....
Large language models (LLMs) transfer well to new tasks out-of-the-box simply given a natural prompt that demonstrates how perform the task and no additional training. Prompting is brittle process wherein small modifications can cause large variations in model predictions, therefore significant effort dedicated towards designing painstakingly "perfect prompt" for task. To mitigate high degree of involved prompt-design, we instead ask whether producing multiple effective, yet imperfect,...
Abstract Background Analyses of Fgf10 and Fgfr2b mutant mice, as well human studies, suggest that FGF10/FGFR2b signaling may play an essential, nonredundant role during embryonic SMG development. To address this question, we have analyzed the phenotype in heterozygous null mice. In addition, although previous studies FGF8/FGFR2c pathways are functionally interrelated, little is known about functional relationship between these two We designed vivo vitro experiments to question. Results mice...
Abstract Fibroblast growth factor (FGF) signaling and β-catenin activation have been shown to be crucial for early embryonic liver development. This study determined the significance of FGF10-mediated in a murine progenitor cell population as well its relation activation. We observed that Fgf10 −/− Fgfr2b mouse livers are smaller than wild-type livers; exhibit diminished proliferation hepatoblasts. A comparison β-galactosidase activity readout expression +/LacZ mice TOPGAL mice, demonstrated...
Little is known about the regulation of cell fate decisions that lead to formation five pairs mammary placodes in surface ectoderm mouse embryo. We have previously shown fibroblast growth factor 10(FGF10) required for 1, 2, 3 and 5. Here,we found Fgf10 expressed only somites underlying 2 3, gradients across within these somites. To test whether somitic FGF10 two placodes, we analyzed a number mutants with different perturbations presence WNT signals ectodermal multilayering, markers line...
Tissue-engineered small intestine (TESI) has successfully been used to rescue Lewis rats after massive bowel resection. In this study, we transitioned the technique a mouse model, allowing investigation of processes involved during TESI formation through transgenic tools available in species. This is necessary step toward applying human therapy. Multicellular organoid units were derived from intestines mice and transplanted within abdomen on biodegradable polymers. Immunofluorescence...
During embryonic development, appropriate dorsoventral patterning of the trachea leads to formation periodic cartilage rings from ventral mesenchyme and continuous smooth muscle dorsal mesenchyme. In this work, we have investigated role two crucial morphogens, fibroblast growth factor 10 sonic hedgehog, in periodically alternating cartilaginous non-cartilaginous domains Using a combination gain- loss-of-function approaches for FGF10 SHH, demonstrate that precise spatio-temporal patterns...
Canonical Wnt signaling plays multiple roles in lung organogenesis and repair by regulating early progenitor cell fates: investigation has been enhanced canonical reporter mice, TOPGAL, BATGAL Axin2LacZ. Although widely used, it remains unclear whether these reporters convey the same information about signaling. We therefore compared beta-galactosidase expression patterns of mice whole embryo versus isolated prenatal lungs. To determine if varied further during repair, we analyzed...
This paper studies problems in data reconstruction, an important area with numerous applications. In particular, we examine the reconstruction of binary and nonbinary sequences from synchronization (insertion/deletion-correcting) codes. These have been corrupted by a fixed number symbol insertions (larger than minimum edit distance code), yielding distinct traces to be used for reconstruction. We wish know needed exact is general version problem tackled Levenshtein uncoded sequences....
In this paper, we investigate the problem of constructing codes capable correcting two deletions. particular, construct a code that requires redundancy approximately 8 log <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> n + O(log n) bits redundancy, where denotes length code. To best authors' knowledge, represents known construction in it lowest number redundant for
In non-volatile memories, reading stored data is typically done through the use of predetermined fixed thresholds. However, due to problems commonly affecting such including voltage drift, overwriting, and inter-cell coupling, threshold usage often results in significant asymmetric errors. To combat these problems, Zhou, Jiang, Bruck recently introduced notion dynamic thresholds applied them binary sequences. this paper, we explore for multi-level cell (MLC) memories. We provide a general...