381 research outputs found
Parsing Science - Semantic Meaning in Images
No full textA picture may be worth 1000 words, but can we also teach computers to create stories from the stories that lie inside our images? In this episode, Devi Parikh of Georgia Tech’s school of interactive computing discusses her work training computers to determine the semantic meaning within images. Devi talks about the stories behind her article "Bringing Semantics Into Focus Using Visual Abstraction," which she co-authored with Larry Zitnick in 2013
Ovarian steroid cell tumor in pregnancy-a rare occurrence: Report of a case and review of the literature
No full textBackground: Steroid-cell tumors of the ovary are very rare, especially during pregnancy, and they must be distinguished from luteoma of pregnancy.
Case: An 18-year-old female, gravida 3, para 1-0-1-1, at 38 weeks' of gestation, had an adnexal mass that was discovered incidentally during a Caesarean section. The tumor was excised and her male infant was normal.
Results: Histologic workup revealed the tumor to be a steroid-cell tumor, which is exceedingly rare in pregnancy.
Conclusions: Ovarian steroid-cell tumors, which are malignant one-third of the time, are difficult to distinguish from luteoma of pregnancy.Peer reviewe
Gender and climate change framework for analysis, policy & action
No full text"This paper by Jyoti Parikh provides a framework to anlayse gender and climate change concerns keeping in view the strengths and vulnerability of poor ‐ women in particular. The author also provides policy recommendations for policies and actions.
The Neuropeptide VGF is Reduced in Human Bipolar Postmortem Brain and Contributes to Some of the Behavioral and Molecular Effects of Lithium
No full textRecent studies demonstrate that the neuropeptide VGF (non-acronymic) is regulated in the hippocampus by antidepressant therapies and animal models of depression and that acute VGF treatment has antidepressant-like activity in animal paradigms. However, the role of VGF in human psychiatric disorders is unknown. We now demonstrate using in situ hybridization that VGF is downregulated in bipolar disorder in the CA region of the hippocampus and Brodmann’s Area 9 (BA9) of the prefrontal cortex. The mechanism of VGF in relation to LiCl was explored. Both LiCl intraperitoneally (IP) and VGF intracerebroventricularly (ICV) reduced latency to drink in novelty-induced hypophagia and LiCl was not effective in VGF+/- mice suggesting that VGF may contribute to the effects of LiCl in this behavioral procedure that responds to chronic antidepressant treatment. VGF by intrahippocampal injection also had novel activity in an amphetamineinduced hyperlocomotion assay thus mimicking the actions of LiCl injected IP in a system that phenocopies manic-like behavior. Moreover, VGF+/- mice exhibited increased locomotion following amphetamine and did not respond to LiCl, suggesting that VGF is required for the effects of LiCl in curbing the response to amphetamine. Finally, VGF by ICV in vivo activated the same signaling pathways as LiCl and is necessary for the induction of MAPK and AKT by LiCl thus lending insight into the molecular mechanisms underlying the actions of VGF. The dysregulation of VGF in bipolar disorder as well as the behavioral effects of the neuropeptide similar to LiCl suggests that VGF may underlie the pathophysiology of bipolar disorder.Peer reviewe
Assessing the effect of exercise on dial-task cognitive impairment in patients with Parkinson's disease
No full textThis work was produced while the author was an undergraduate student in the Summer Research Institute of the Ronald E. McNair Post Baccalaureate Degree Achievement Program at Rutgers University
Examining mechanical properties of single acetaminophen crystal using nanoindentation methods
No full textThe pharmaceutical industry incurs substantial loss in revenue and consumer confidence with inefficient manufacturing practices. Large scale processing of organic compounds is challenging due to its sensitivity to environmental conditions and the unpredictable breakage behavior of tablets under applied stress. Tablet compaction and particle size reduction through milling induces variability in the end product. Variability in powder flow, stress induced transformation in polymorphic compounds, re-crystallization after compaction, and lack of content uniformity are some factors that translate into poor product quality. These challenges can be partially resolved by a better understanding of mechanical properties of crystalline pharmaceutical materials at single particle level. The endeavor of this study was to understand the breakage behavior of various planes of a single Acetaminophen crystal using nanoindentation instrumentation. The results of the study indicated that the Acetaminophen crystal is anisotropic with respect to hardness and Young’s modulus values. Analysis of the load-depth curve, discontinuities on the loading and unloading cycle were observed, as well as pop-in events during constant load intervals. Furthermore, the frequency of pop-in events on the loading depth curve was found to correlate with the elasticity of the planes in question. It was also apparent that the organic compound was sensitive to environmental conditions. Varying strain rates effects different planes of the same crystal and also in adhesion reflected sensitivity to environmental conditions. The exact mechanism by which the crystal deforms is still unknown. However it is theorized that it could be through partial dislocations and crack propagations.M.S.Includes bibliographical referencesby Hiral Parik
Learning visual tasks with selective attention
No full textKnowing where to look in an image can significantly improve performance in computer vision tasks by eliminating irrelevant information from the rest of the input image, and by breaking down complex scenes into simpler and more familiar sub-components. We show that a framework for identifying multiple task-relevant regions can be learned in current state-of-the-art deep network architectures, resulting in significant gains in several visual prediction tasks. We will demonstrate both directly and indirectly supervised models for selecting image regions and show how they can improve performance over baselines by means of focusing on the right areas.Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2017-09-29 without embargo termsThe student, Kevin Shih, accepted the attached license on 2017-07-10 at 12:45.The student, Kevin Shih, submitted this Dissertation for approval on 2017-07-10 at 13:18.This Dissertation was approved for publication on 2017-07-11 at 15:03.DSpace SAF Submission Ingestion Package generated from Vireo submission #11368 on 2017-09-29 at 11:29:29Made available in DSpace on 2017-09-29T17:56:40Z (GMT). No. of bitstreams: 2
SHIH-DISSERTATION-2017.pdf: 35992565 bytes, checksum: 0236e3afe4b94ec89729250662a7eb76 (MD5)
LICENSE.txt: 4207 bytes, checksum: 850b64b383db31c4fc6d39801a3eab05 (MD5)
Previous issue date: 2017-07-1
Role of Premises in Visual Question Answering
No full textIn this work, we make a simple but important observation questions about images often contain premises -- objects and relationships implied by the question -- and that reasoning about premises can help Visual Question Answering (VQA) models respond more intelligently to irrelevant or previously unseen questions.
When presented with a question that is irrelevant to an image, state-of-the-art VQA models will still answer based purely on learned language biases, resulting in nonsensical or even misleading answers. We note that a visual question is irrelevant to an image if at least one of its premises is false (i.e. not depicted in the image). We leverage this observation to construct a dataset for Question Relevance Prediction and Explanation (QRPE) by searching for false premises. We train novel irrelevant question detection models and show that models that reason about premises consistently outperform models that do not.
We also find that forcing standard VQA models to reason about premises during training can lead to improvements on tasks requiring compositional reasoning.Master of ScienceThere has been substantial recent work on the Visual Question Answering (VQA) problem in which an automated agent is tasked on answering questions about images posed in natural language. In this work, we make a simple but important observation – questions about images often contain premises – objects and relationships implied by the question – and that reasoning about premises can help VQA models respond more intelligently to irrelevant or previously unseen questions.
When presented with a question that is irrelevant to an image, state-of-the-art VQA models will still answer based purely on learned language biases, resulting in nonsensical or even misleading answers. We note that a visual question is irrelevant to an image if at least one of its premises is false (i.e. not depicted in the image). We leverage this observation to construct a dataset for Question Relevance Prediction and Explanation (QRPE) by searching for false premises. We train novel irrelevant question detection models and show that models that reason about premises consistently outperform models that do not.
We also find that forcing standard VQA models to reason about premises during training can lead to improvements on tasks requiring compositional reasoning
Recommended from our members
Computational fashion understanding
No full textFashion is very much a social and cultural statement. It documents the tastes and values of an era, and can even reflect the political orientation and economic development of a nation. While a fashion style can be defined by combinations of basic elements, fashion itself can take on indefinite forms. It is shaped by group acceptance, social forces important in an era, people's desire to relate to specific lifestyles, and it evolves over time. Fashion being so tightly wrapped up in our daily lives, our fashion appearance becomes one of the most direct ways to express ourselves.
Understanding fashion computationally would allow better personalization, help the industry analyze its enterprise, and bring us closer towards building socially intelligent agents. Unfortunately, current computer vision systems for fashion are lacking in multiple ways. They fail to consider the synergy between multiple garments that compose an outfit, the garments' relation to the human wearer, and the underlying social factors that influence our fashion choices.
The goal of this thesis is to devise computational models that understand fashion in the full context of outfits, humans and society. Our solution is to develop models that learn to utilize "in-the-wild" fashion images with their associated meta-data (e.g., geotags, timestamps, user tags), and capture human's aesthetics and rationales for fashion in an unsupervised manner. These freely available Internet images come at large-scale and are natural examples of what people choose to wear, allowing models with easy adaptation to new definitions of fashion as it evolves.
The thesis begins by introducing our method that accounts for the synergy between multiple garments in an outfit. Specifically, we devise a topic-model-based method that automatically analyzes the interplay between garments, and discovers underlying combinations that form fashion styles. As outfits are eventually worn by humans, the thesis moves on to introduce new fashion recommendation systems that are aware of an individual's body shape and give suggestions on dressing in a more fashionable way. A 3D mesh of a human body is recovered and mapped into a visual embedding to identify flattering garments for the given body shape. In order to communicate with users how to dress fashionably, our model learns to score fashionability of outfits by seeing tens of thousands of natural examples from Internet images, and renders the optimized look of an outfit via an image generation neural network. Finally, the thesis zooms out to consider fashion in the history of human culture, where we conduct our study on millions of news articles and fashion images spanning from the beginning of the twentieth century to nowadays. We first extend our style topic model to adapt to the fluid and ever-evolving nature of fashion styles, and then devise a multi-modal statistical model to identify the links from world events to fashion evolution. All together, I hope my research lays down foundations for human-aware, society-aware, and evolving fashion understanding.Computer Scienc
- …
