Filters








3,918 Hits in 2.3 sec

Possible Lyme Carditis with Sick Sinus Syndrome

Brian Cheung, Larry Lutwick, Michelle Cheung
2020 IDCases  
Borrelia burgdorferi (B. burgdorferi) is a spirochete bacterium that is transmitted via the Ixodes tick. Infection results in Lyme disease with possible cardiac manifestations, which is also known as Lyme carditis. Patients can present with bradycardia due to rapidly fluctuating atrioventricular block (AVB), which is the hallmark of Lyme carditis. However, we present a rare case of sick sinus syndrome (SSS) without AVB in a 47-year-old man with Lyme disease. He initially presented with a
more » ... e and subsequently developed new onset bradycardia and a right cranial nerve (CN) VI palsy with diplopia. B. burgdorferi enzyme-linked immunosorbent assay (ELISA) screen and IgM western blot were positive. He was admitted to the intensive care unit. Electrocardiography (EKG) indicated a heart rate in the high 30 s beats per minute (BPM) with several pauses, but no AVB was present. The patient responded well to therapy, and was discharged with an outpatient regimen of doxycycline. Lyme carditis should be considered in patients who develop new onset bradycardia and live in endemic areas.
doi:10.1016/j.idcr.2020.e00761 pmid:32368492 pmcid:PMC7190756 fatcat:kcjh6262enfj5ndg2jmn3vdneu

Recurrent Parameter Generators [article]

Jiayun Wang, Yubei Chen, Stella X. Yu, Brian Cheung, Yann LeCun
2021 arXiv   pre-print
We present a generic method for recurrently using the same parameters for many different convolution layers to build a deep network. Specifically, for a network, we create a recurrent parameter generator (RPG), from which the parameters of each convolution layer are generated. Though using recurrent models to build a deep convolutional neural network (CNN) is not entirely new, our method achieves significant performance gain compared to the existing works. We demonstrate how to build a
more » ... neural network to achieve similar performance compared to other traditional CNN models on various applications and datasets. Such a method allows us to build an arbitrarily complex neural network with any amount of parameters. For example, we build a ResNet34 with model parameters reduced by more than 400 times, which still achieves 41.6% ImageNet top-1 accuracy. Furthermore, we demonstrate the RPG can be applied at different scales, such as layers, blocks, or even sub-networks. Specifically, we use the RPG to build a ResNet18 network with the number of weights equivalent to one convolutional layer of a conventional ResNet and show this model can achieve 67.2% ImageNet top-1 accuracy. The proposed method can be viewed as an inverse approach to model compression. Rather than removing the unused parameters from a large model, it aims to squeeze more information into a small number of parameters. Extensive experiment results are provided to demonstrate the power of the proposed recurrent parameter generator.
arXiv:2107.07110v1 fatcat:lq35rnwurjcjlcnk3tlem37dm4

Fast Simultaneous Training of Generalized Linear Models (FaSTGLZ) [article]

Bryan R. Conroy, Jennifer M. Walz, Brian Cheung, Paul Sajda
2013 arXiv   pre-print
We present an efficient algorithm for simultaneously training sparse generalized linear models across many related problems, which may arise from bootstrapping, cross-validation and nonparametric permutation testing. Our approach leverages the redundancies across problems to obtain significant computational improvements relative to solving the problems sequentially by a conventional algorithm. We demonstrate our fast simultaneous training of generalized linear models (FaSTGLZ) algorithm on a
more » ... ber of real-world datasets, and we run otherwise computationally intensive bootstrapping and permutation test analyses that are typically necessary for obtaining statistically rigorous classification results and meaningful interpretation. Code is freely available at http://liinc.bme.columbia.edu/fastglz.
arXiv:1307.8430v1 fatcat:cw2pya5nlraxthe6jiequ7tyka

Equivariant Contrastive Learning [article]

Rumen Dangovski, Li Jing, Charlotte Loh, Seungwook Han, Akash Srivastava, Brian Cheung, Pulkit Agrawal, Marin Soljačić
2022 arXiv   pre-print
In state-of-the-art self-supervised learning (SSL) pre-training produces semantically good representations by encouraging them to be invariant under meaningful transformations prescribed from human knowledge. In fact, the property of invariance is a trivial instance of a broader class called equivariance, which can be intuitively understood as the property that representations transform according to the way the inputs transform. Here, we show that rather than using only invariance, pre-training
more » ... that encourages non-trivial equivariance to some transformations, while maintaining invariance to other transformations, can be used to improve the semantic quality of representations. Specifically, we extend popular SSL methods to a more general framework which we name Equivariant Self-Supervised Learning (E-SSL). In E-SSL, a simple additional pre-training objective encourages equivariance by predicting the transformations applied to the input. We demonstrate E-SSL's effectiveness empirically on several popular computer vision benchmarks, e.g. improving SimCLR to 72.5% linear probe accuracy on ImageNet. Furthermore, we demonstrate usefulness of E-SSL for applications beyond computer vision; in particular, we show its utility on regression problems in photonics science. Our code, datasets and pre-trained models are available at https://github.com/rdangovs/essl to aid further research in E-SSL.
arXiv:2111.00899v2 fatcat:xc5nasvgmnhhzgpr4hnvszqz3a

Anti-GD2 antibody 3F8 and barley-derived (1 → 3),(1 → 4)-β-D-glucan

Shakeel Modak, Brian H. Kushner, Kim Kramer, Andrew Vickers, Irene Y. Cheung, Nai-Kong V. Cheung
2013 Oncoimmunology  
doi:10.4161/onci.23402 pmid:23802080 pmcid:PMC3661165 fatcat:bzfaa6oehzgmzl4l7h2pbmjyja

Cautious Adaptation For Reinforcement Learning in Safety-Critical Settings [article]

Jesse Zhang, Brian Cheung, Chelsea Finn, Sergey Levine, Dinesh Jayaraman
2020 arXiv   pre-print
Reinforcement learning (RL) in real-world safety-critical target settings like urban driving is hazardous, imperiling the RL agent, other agents, and the environment. To overcome this difficulty, we propose a "safety-critical adaptation" task setting: an agent first trains in non-safety-critical "source" environments such as in a simulator, before it adapts to the target environment where failures carry heavy costs. We propose a solution approach, CARL, that builds on the intuition that prior
more » ... perience in diverse environments equips an agent to estimate risk, which in turn enables relative safety through risk-averse, cautious adaptation. CARL first employs model-based RL to train a probabilistic model to capture uncertainty about transition dynamics and catastrophic states across varied source environments. Then, when exploring a new safety-critical environment with unknown dynamics, the CARL agent plans to avoid actions that could lead to catastrophic states. In experiments on car driving, cartpole balancing, half-cheetah locomotion, and robotic object manipulation, CARL successfully acquires cautious exploration behaviors, yielding higher rewards with fewer failures than strong RL adaptation baselines. Website at https://sites.google.com/berkeley.edu/carl.
arXiv:2008.06622v1 fatcat:obkgm3b5trfzrdbtyh4rjasc2a

Word Embedding Visualization Via Dictionary Learning [article]

Juexiao Zhang, Yubei Chen, Brian Cheung, Bruno A Olshausen
2021 arXiv   pre-print
Alona Fyshe, Partha P Talukdar, Brian Murphy, and Tom M Mitchell. 2014. Interpretable semantic vectors from a joint model of brain-and text-based meaning. In Proceedings of the conference.  ...  Brian Murphy, Partha Talukdar, and Tom Mitchell. 2012. Learning effective and interpretable semantic models using non-negative sparse embedding. In Proceedings of COLING 2012, pages 1933-1950.  ... 
arXiv:1910.03833v2 fatcat:lzwbcoex3zbw3aigew5ltd4sjy

Learning to Learn Without Labels

Luke Metz, Niru Maheswaranathan, Brian Cheung, Jascha Sohl-Dickstein
2018 International Conference on Learning Representations  
dblp:conf/iclr/MetzMCS18 fatcat:5i3i3rerlrbajphbyiyo3i56ky

Superposition of many models into one [article]

Brian Cheung, Alex Terekhov, Yubei Chen, Pulkit Agrawal, Bruno Olshausen
2019 arXiv   pre-print
Correspondence to: Brian Cheung <bcheung@berkeley.edu>. tasks, individual models for each task can coexist with one another in superposition.  ... 
arXiv:1902.05522v2 fatcat:e5u45dm6vfberlidfpujtcp52a

Salience Bias in Crowdsourcing Contests

Ho Cheung Brian Lee, Sulin Ba, Xinxin Li, Jan Stallaert
2018 Information systems research  
Crowdsourcing relies on online platforms to connect a community of users to perform specific tasks. However, without appropriate control, the behavior of the online community might not align with the platform's designed objective, which can lead to an inferior platform performance. This paper investigates how the feedback information on a crowdsourcing platform and systematic bias of crowdsourcing workers can affect crowdsourcing outcomes. Specifically, using archival data from the online
more » ... ourcing platform Kaggle, combined with survey data from actual Kaggle contest participants, we examine the role of a systematic bias, namely the salience bias, in influencing the performance of the crowdsourcing workers and how the number of crowdsourcing workers moderates the impact of the salience bias as a result of the parallel path effect and competition effect. Our results suggest that the salience bias influences the performance of contestants, including the winners of the contests. Furthermore, the parallel path effect cannot completely eliminate the impact of the salience bias, but it can attenuate it to a certain extent. By contrast, the competition effect is likely to amplify the impact of the salience bias. Our results have critical implications for crowdsourcing firms and platform designers.
doi:10.1287/isre.2018.0775 fatcat:qk2p7sgsejgqbebj2jn7sw4v2i

Early Molecular Response of Marrow Disease to Biologic Therapy Is Highly Prognostic in Neuroblastoma

Irene Y. Cheung, M. Serena Lo Piccolo, Brian H. Kushner, Nai-Kong V. Cheung
2003 Journal of Clinical Oncology  
Cheung, M. Serena Lo Piccolo, Brian H. Kushner, and Nai-Kong V.  ...  Lo Piccolo MS, Cheung NKV, Cheung IY: GD2 Synthase: A new molecular marker for detecting neuroblastoma. Cancer 92:924-931, 2001 13.  ... 
doi:10.1200/jco.2003.11.077 pmid:14551304 fatcat:drzjetyownf2pcjm4pxopezya4

Safety of statins: an update

Miao Hu, Bernard M.Y. Cheung, Brian Tomlinson
2012 Therapeutic Advances in Drug Safety  
Professor Brian Tomlinson has received research funding to perform clinical studies from AstraZeneca, Bayer, Boehringer Ingelheim, Daiichi Sankyo, Kowa, Merck Serono, Merck Sharp and Dohme, Novartis, Otsuka  ... 
doi:10.1177/2042098612439884 pmid:25083232 pmcid:PMC4110822 fatcat:y22sldvsufggba6cecgc47rrcq

GRIBCG: A software for selection of sgRNAs in the design of balancer chromosomes [article]

Brian B Merritt, Lily S Cheung
2018 bioRxiv   pre-print
Balancer chromosomes are tools used by fruit fly geneticists to prevent meiotic recombination. Recently, CRISPR/Cas9 genome editing has been shown capable of generating inversions similar to the chromosomal rearrangements present in balancer chromosomes. Extending the benefits of balancer chromosomes to other multicellular organisms could significantly accelerate biomedical and plant genetics research. Results: Here, we present GRIBCG (Guide RNA Identifier for Balancer Chromosome Generation), a
more » ... tool for the rational design of balancer chromosomes. GRIBCG identifies single guide RNAs (sgRNAs) for use with Streptococcus pyogenes Cas9 (SpCas9). These sgRNAs would efficiently cut a chromosome multiple times while minimizing off-target cutting in the rest of the genome. We describe the performance of this tool on six model organisms and compare our results to two routinely used fruit fly balancer chromosomes. Conclusion: GRIBCG is the first of its kind tool for the design of balancer chromosomes using CRISPR/Cas9. GRIBCG can accelerate genetics research by providing a fast, systematic and simple to use framework to induce chromosomal rearrangements.
doi:10.1101/484360 fatcat:mbipnyhl7rc4jpju2mld34axqa

Parameters affecting scalable underwater compressed air energy storage

Brian C. Cheung, Rupp Carriveau, David S.-K. Ting
2014 Applied Energy  
h i g h l i g h t s UWCAES design parameters were studied to determine influence on round-trip efficiency and exergy destruction. Air compression and expansion contributed the most to total system exergy destruction for all parametric study cases. The system was most sensitive to pipe diameter, followed by expander and compressor efficiencies, and air storage depth. Increasing expander and compressor efficiencies showed greatest improvements to UWCAES performance. a b s t r a c t Underwater
more » ... ressed air energy storage (UWCAES) is founded on mature concepts, many of them sourced from underground compressed air energy storage technology. A fundamental difference between the two systems is the way in which air is stored. UWCAES utilizes distensible boundary, submerged air accumulators as opposed to rigid walled caverns. This paper presents an analysis of the primary design parameters in a basic UWCAES system. The results from the parametric study and first-order sensitivity analysis show the importance and impact each design parameter has on overall system performance and can serve as a first reference guideline in system design. The analysis revealed significant system sensitivities to pipe diameter, expander and compressor efficiencies, and air storage depth. The air compression and expansion processes contributed most to system exergy destruction for all parametric study cases.
doi:10.1016/j.apenergy.2014.08.028 fatcat:n7eotaztpfar3agwkrl6poggca

In Reply:

Brian H. Kushner, Kim Kramer, Michael P. Laquaglia, Shakeel Modak, Karima Yataghene, Nai-Kong V. Cheung
2005 Journal of Clinical Oncology  
Brian H. Kushner, Kim Kramer, Michael P. Laquaglia, Shakeel Modak, Karima Yataghene, and Nai-Kong V.  ...  Cheung Authors’ Disclosures of Potential Conflicts of Interest he authors indicated no potential conflicts of interest. REFERENCES 1 Should We Transplant Indolent Lymphoma?  ... 
doi:10.1200/jco.2005.01.7673 fatcat:ea55yvohe5cmhb376u7bbysse4
« Previous Showing results 1 — 15 out of 3,918 results