Filters








49,502 Hits in 1.5 sec

Corporatism Reconsidered: Howard J. Wiarda's Legacy

Linda Chen
2018 Polity  
I n the 1960s, as Howard J. Wiarda finished up his graduate studies at the University of Florida, Latin America was in turmoil.  ...  Wiarda, "Toward a Framework," 208-17 (see note 2 above). 7 . 7 Ibid., 206-35; Howard J. Wiarda, Latin American Politics (New York: Wadsworth, 1995), ch. 1. 8.  ... 
doi:10.1086/699630 fatcat:myw7gtftmvbnvgon2gdk6vqi7m

Can Rationalization Improve Robustness? [article]

Howard Chen, Jacqueline He, Karthik Narasimhan, Danqi Chen
2022 arXiv   pre-print
A growing line of work has investigated the development of neural NLP models that can produce rationales--subsets of input that can explain their model predictions. In this paper, we ask whether such rationale models can also provide robustness to adversarial attacks in addition to their interpretable nature. Since these models need to first generate rationales ("rationalizer") before making predictions ("predictor"), they have the potential to ignore noise or adversarially added text by simply
more » ... masking it out of the generated rationale. To this end, we systematically generate various types of 'AddText' attacks for both token and sentence-level rationalization tasks, and perform an extensive empirical evaluation of state-of-the-art rationale models across five different tasks. Our experiments reveal that the rationale models show the promise to improve robustness, while they struggle in certain scenarios--when the rationalizer is sensitive to positional bias or lexical choices of attack text. Further, leveraging human rationale as supervision does not always translate to better performance. Our study is a first step towards exploring the interplay between interpretability and robustness in the rationalize-then-predict framework.
arXiv:2204.11790v2 fatcat:yhslidfo25herlpzejp5rpelhy

Intercultural communication competence: Some perspectives of research

Guo‐Ming Chen
1990 The Howard Journal of Communications  
Research about components of ICC can be roughly examined from four perspectives: personality strength, communication skills, psychological adaptation, and cultural awareness (Abe & Wiseman, 1983; Chen  ...  Studies from Abe and Wiseman (1983) , Chen (1989), Hammer, Gudykunst, and Wiseman (1978) . and Martin (1987) have indicated that to be competent in intercultural interaction one has to understand the  ... 
doi:10.1080/10646179009359718 fatcat:yzqde64fwjf2xg4s5wu6gt7you

Learning to Optimize: A Primer and A Benchmark [article]

Tianlong Chen, Xiaohan Chen, Wuyang Chen, Howard Heaton, Jialin Liu, Zhangyang Wang, Wotao Yin
2021 arXiv   pre-print
Chen, X. Chen, W. Chen, H. Heaton, J. Liu, Z. Wang and W. Yin . Methods that are underscored in this section are also later evaluated and benchmarked in Section. 4. .  ... 
arXiv:2103.12828v2 fatcat:c75y3wz6cngirb2zpugjk63ymq

Non-Parametric Few-Shot Learning for Word Sense Disambiguation [article]

Howard Chen, Mengzhou Xia, Danqi Chen
2021 arXiv   pre-print
Word sense disambiguation (WSD) is a long-standing problem in natural language processing. One significant challenge in supervised all-words WSD is to classify among senses for a majority of words that lie in the long-tail distribution. For instance, 84% of the annotated words have less than 10 examples in the SemCor training data. This issue is more pronounced as the imbalance occurs in both word and sense distributions. In this work, we propose MetricWSD, a non-parametric few-shot learning
more » ... roach to mitigate this data imbalance issue. By learning to compute distances among the senses of a given word through episodic training, MetricWSD transfers knowledge (a learned metric space) from high-frequency words to infrequent ones. MetricWSD constructs the training episodes tailored to word frequencies and explicitly addresses the problem of the skewed distribution, as opposed to mixing all the words trained with parametric models in previous work. Without resorting to any lexical resources, MetricWSD obtains strong performance against parametric alternatives, achieving a 75.1 F1 score on the unified WSD evaluation benchmark (Raganato et al., 2017b). Our analysis further validates that infrequent words and senses enjoy significant improvement.
arXiv:2104.12677v2 fatcat:6hpfv47pxrh7tdscwyd5coljlu

Comparative manufacture and cell-based delivery of antiretroviral nanoformulations

Howard Gendelman, Balkundi, Nowacek, Veerhubhotla, Chen, Martinez-Skinner, Roy, Mosley, Kanmogne, Alexander Kabanov, Bronich, McMillan (+1 others)
2011 International Journal of Nanomedicine  
doi:10.2147/ijn.s27830 pmid:22267924 pmcid:PMC3260033 fatcat:cnnqhhzahffwbmquinohmeepne

Searching for MobileNetV3 [article]

Andrew Howard, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang, Yukun Zhu, Ruoming Pang, Vijay Vasudevan, Quoc V. Le, Hartwig Adam
2019 arXiv   pre-print
We present the next generation of MobileNets based on a combination of complementary search techniques as well as a novel architecture design. MobileNetV3 is tuned to mobile phone CPUs through a combination of hardware-aware network architecture search (NAS) complemented by the NetAdapt algorithm and then subsequently improved through novel architecture advances. This paper starts the exploration of how automated search algorithms and network design can work together to harness complementary
more » ... roaches improving the overall state of the art. Through this process we create two new MobileNet models for release: MobileNetV3-Large and MobileNetV3-Small which are targeted for high and low resource use cases. These models are then adapted and applied to the tasks of object detection and semantic segmentation. For the task of semantic segmentation (or any dense pixel prediction), we propose a new efficient segmentation decoder Lite Reduced Atrous Spatial Pyramid Pooling (LR-ASPP). We achieve new state of the art results for mobile classification, detection and segmentation. MobileNetV3-Large is 3.2\% more accurate on ImageNet classification while reducing latency by 15\% compared to MobileNetV2. MobileNetV3-Small is 4.6\% more accurate while reducing latency by 5\% compared to MobileNetV2. MobileNetV3-Large detection is 25\% faster at roughly the same accuracy as MobileNetV2 on COCO detection. MobileNetV3-Large LR-ASPP is 30\% faster than MobileNetV2 R-ASPP at similar accuracy for Cityscapes segmentation.
arXiv:1905.02244v4 fatcat:to3jew7bynho3k2yx6tbhb6era

Towards Federated Long-Tailed Learning [article]

Zihan Chen, Songshang Liu, Hualiang Wang, Howard H. Yang, Tony Q.S. Quek, Zuozhu Liu
2022 arXiv   pre-print
Data privacy and class imbalance are the norm rather than the exception in many machine learning tasks. Recent attempts have been launched to, on one side, address the problem of learning from pervasive private data, and on the other side, learn from long-tailed data. However, both assumptions might hold in practical applications, while an effective method to simultaneously alleviate both issues is yet under development. In this paper, we focus on learning with long-tailed (LT) data
more » ... s under the context of the popular privacy-preserved federated learning (FL) framework. We characterize three scenarios with different local or global long-tailed data distributions in the FL framework, and highlight the corresponding challenges. The preliminary results under different scenarios reveal that substantial future work are of high necessity to better resolve the characterized federated long-tailed learning tasks.
arXiv:2206.14988v1 fatcat:orr64badojeejgeixuunrjssdq

Safeguarded Learned Convex Optimization [article]

Howard Heaton and Xiaohan Chen and Zhangyang Wang and Wotao Yin
2020 arXiv   pre-print
Many applications require repeatedly solving a certain type of optimization problem, each time with new (but similar) data. Data-driven algorithms can "learn to optimize" (L2O) with much fewer iterations and with similar cost per iteration as general-purpose optimization algorithms. L2O algorithms are often derived from general-purpose algorithms, but with the inclusion of (possibly many) tunable parameters. Exceptional performance has been demonstrated when the parameters are optimized for a
more » ... rticular distribution of data. Unfortunately, it is impossible to ensure all L2O algorithms always converge to a solution. However, we present a framework that uses L2O updates together with a safeguard to guarantee convergence for convex problems with proximal and/or gradient oracles. The safeguard is simple and computationally cheap to implement, and it should be activated only when the current L2O updates would perform poorly or appear to diverge. This approach yields the numerical benefits of employing machine learning methods to create rapid L2O algorithms while still guaranteeing convergence. Our numerical examples demonstrate the efficacy of this approach for existing and new L2O schemes.
arXiv:2003.01880v2 fatcat:nkn2u7epvfbilprcu6t6sgwiti

PEOPLE IDENTIFICATION ACROSS AMBIENT CAMERA NETWORKS [chapter]

DATONG CHEN, ASHOK BHARUSHA, HOWARD WACTLAR
2007 Information Sciences 2007  
This paper proposes a robust algorithm based on conditional random field (CRF) to improve people identification in an ambient camera network environment. Ambient cameras collect visual signals of subjects from a distance. The quality of these visual signals are often not adequate to train effective classifiers to identify persons in video. We use a CRF to probabilistically propagate the weak identification results across the temporal-spatial space covered by the camera network. No visual based
more » ... racking is necessary for our algorithm. Experiments with a 23-camera network demonstrate that the proposed algorithm leverages the advantage of monitoring from multiple cameras and improves the accuracy of person identification.
doi:10.1142/9789812709677_0214 fatcat:giqrntwgsfgrdbxisxqh7sekiy

Learning A Minimax Optimizer: A Pilot Study

Jiayi Shen, Xiaohan Chen, Howard Heaton, Tianlong Chen, Jialin Liu, Wotao Yin, Zhangyang Wang
2021 International Conference on Learning Representations  
(Chen et al., 2017) leverages RNN to train a meta-optimizer to optimize black-box functions.  ...  Previously it was also found effective in L2O for minimization problems (Chen et al., 2020a) .  ... 
dblp:conf/iclr/ShenCHC0YW21 fatcat:fikryxn2unherjegjdvo4ffqb4

RecT recombinase expression enables efficient gene editing in Enterococcus [article]

Victor Chen, Matthew E Griffin, Howard C Hang
2020 bioRxiv   pre-print
Enterococcus faecium is a ubiquitous Gram-positive bacterium that has been recovered from the environment, food and microbiota of mammals. Commensal strains of E. faecium can confer beneficial effects on host physiology and immunity, but antibiotic usage has afforded antibiotic-resistant and pathogenic isolates from livestock and humans. However, the dissection of E. faecium functions and mechanisms has been restricted by inefficient gene editing methods. To address these limitations, here we
more » ... port the expression of E. faecium RecT recombinase significantly improves the efficiency of recombineering technologies in commensal strains of E. faecium and other Enterococcus species such as E. durans and E. hirae. Notably, we demonstrate that E. faecium RecT expression facilitated the chromosomal insertion of both single-stranded and double-stranded DNA templates encoding antibiotic selectable markers. Moreover, the expression of RecT in combination with clustered regularly interspaced palindromic repeat (CRISPR)-Cas9 and guide RNAs (gRNAs) enabled highly efficient scar-less ssDNA recombineering to generate specific gene editing mutants in E. faecium. The RecT-mediated recombineering methods described here should significantly enhance genetic studies of E. faecium and other closely related species for functional and mechanistic studies. Importance Enterococcus faecium is widely recognized as an emerging public health threat with the rise of drug resistance and nosocomial infections. Nevertheless, commensal Enterococcus strains possess beneficial health functions in mammals to upregulate host immunity and prevent microbial infections. This functional dichotomy of Enterococcus species and strains highlights the need for in-depth studies to discover and characterize the genetic components underlining its diverse activities. However, genetic engineering in E. faecium still requires passive homologous recombination, which often requires cloning of multiple homologous fragments and screening. To alleviate these challenges, we discovered that RecT-recombinase enables more efficient integration of mutagenic DNA templates to generate insertions, deletions and substitutions of genomic DNA in E. faecium. These improved recombineering methods should facilitate functional and mechanistic studies of Enterococcus.
doi:10.1101/2020.09.01.278044 fatcat:bdy55gl67vam3dztxpiy6jgm2m

Large diffractive/refractive apertures for space and airborne telescopes

Howard A. MacEwen, James B. Breckinridge, Khanh D. Pham, Joseph L. Cox, Richard T. Howard, Genshe Chen
2013 Sensors and Systems for Space Applications VI  
Recent work, specifically the Lawrence Livermore National Laboratory (LLNL) Eyeglass and the DARPA MOIRE programs, have evaluated lightweight, easily packaged and deployed, diffractive/refractive membrane transmissive lenses as entrance apertures for large space and airborne telescopes. This presentation describes a new, innovative approach to the theory of diffractive and refractive effects in lenses used as telescope entrance apertures and the fabrication of the necessary large membrane
more » ... . Analyses are presented to indicate how a broadband, highly transmissive diffractive / refractive membrane lens can be developed and fabricated, and potential applications in defense and astronomy are briefly discussed. * This has been validated with a wavefront (phase-matching) approach (based upon work of Johnson 14 ) shown to achieve the same final results under the same conditions. * Graphs are based upon 10% R 0 decrements for each step. † A slight difference from the 1.5 value used by Meinel and Meinel. ‡ Based upon telescope parameters defined in Figure 5 .
doi:10.1117/12.2015457 fatcat:lad5g6k5wfa3jjnfz7od5uitfi

Serotonergic modulation of walking in Drosophila [article]

Clare E Howard, Chin-Lin Chen, Tanya Tabachnik, Rick Hormigo, Pavan Ramdya, Richard S Mann
2019 bioRxiv   pre-print
To navigate complex environments, animals must generate highly robust, yet flexible, locomotor behaviors. For example, walking speed must be tailored to the needs of a particular environment: Not only must animals choose the correct speed and gait, they must also rapidly adapt to changing conditions, and respond to sudden and surprising new stimuli. Neuromodulators, particularly the small biogenic amine neurotransmitters, allow motor circuits to rapidly alter their output by changing their
more » ... ional connectivity. Here we show that the serotonergic system in the vinegar fly, Drosophila melanogaster, can modulate walking speed in a variety of contexts and in response to sudden changes in the environment. These multifaceted roles of serotonin in locomotion are differentially mediated by a family of serotonergic receptors with distinct activities and expression patterns.
doi:10.1101/753624 fatcat:upjo4ra4avaytjrbugylwvf7oy

Server Free Wireless Federated Learning: Architecture, Algorithm, and Analysis [article]

Howard H. Yang, Zihan Chen, Tony Q. S. Quek
2022 arXiv   pre-print
We demonstrate that merely analog transmissions and match filtering can realize the function of an edge server in federated learning (FL). Therefore, a network with massively distributed user equipments (UEs) can achieve large-scale FL without an edge server. We also develop a training algorithm that allows UEs to continuously perform local computing without being interrupted by the global parameter uploading, which exploits the full potential of UEs' processing power. We derive convergence
more » ... s for the proposed schemes to quantify their training efficiency. The analyses reveal that when the interference obeys a Gaussian distribution, the proposed algorithm retrieves the convergence rate of a server-based FL. But if the interference distribution is heavy-tailed, then the heavier the tail, the slower the algorithm converges. Nonetheless, the system run time can be largely reduced by enabling computation in parallel with communication, whereas the gain is particularly pronounced when communication latency is high. These findings are corroborated via excessive simulations.
arXiv:2204.07609v1 fatcat:dvrs4tcdijb47gmi4e5wv5s62i
« Previous Showing results 1 — 15 out of 49,502 results