A CMOS analog winner-take-all network for large-scale applications

1998 ◽  
Vol 45 (3) ◽  
pp. 300-304 ◽  
Author(s):  
A. Demosthenous ◽  
S. Smedley ◽  
J. Taylor
Keyword(s):  
Large Scale ◽  
Winner Take All ◽  
Take All

scholarly journals Almost Unsupervised Learning for Dense Crowd Counting

2019 ◽  
Vol 33 ◽  
pp. 8868-8875 ◽  
Author(s):  
Deepak Babu Sam ◽  
Neeraj N Sajjan ◽  
Himanshu Maurya ◽  
R. Venkatesh Babu
Keyword(s):  
Unsupervised Learning ◽  
Large Scale ◽  
Learning Method ◽  
Large Variability ◽  
Proposed Model ◽  
Dense Crowd ◽  
Winner Take All ◽  
Learned Features ◽  
Take All

We present an unsupervised learning method for dense crowd count estimation. Marred by large variability in appearance of people and extreme overlap in crowds, enumerating people proves to be a difficult task even for humans. This implies creating large-scale annotated crowd data is expensive and directly takes a toll on the performance of existing CNN based counting models on account of small datasets. Motivated by these challenges, we develop Grid Winner-Take-All (GWTA) autoencoder to learn several layers of useful filters from unlabeled crowd images. Our GWTA approach divides a convolution layer spatially into a grid of cells. Within each cell, only the maximally activated neuron is allowed to update the filter. Almost 99.9% of the parameters of the proposed model are trained without any labeled data while the rest 0.1% are tuned with supervision. The model achieves superior results compared to other unsupervised methods and stays reasonably close to the accuracy of supervised baseline. Furthermore, we present comparisons and analyses regarding the quality of learned features across various models.

scholarly journals Beware the Black-Box: On the Robustness of Recent Defenses to Adversarial Examples

Entropy ◽  
2021 ◽  
Vol 23 (10) ◽  
pp. 1359
Author(s):  
Kaleel Mahmood ◽  
Deniz Gurevin ◽  
Marten van van Dijk ◽  
Phuoung Ha Nguyen
Keyword(s):  
Large Scale ◽  
Error Correcting Codes ◽  
Black Box ◽  
Clear Picture ◽  
Buffer Zones ◽  
Adversarial Models ◽  
Adversarial Examples ◽  
Winner Take All ◽  
The Relationship ◽  
Take All

Many defenses have recently been proposed at venues like NIPS, ICML, ICLR and CVPR. These defenses are mainly focused on mitigating white-box attacks. They do not properly examine black-box attacks. In this paper, we expand upon the analyses of these defenses to include adaptive black-box adversaries. Our evaluation is done on nine defenses including Barrage of Random Transforms, ComDefend, Ensemble Diversity, Feature Distillation, The Odds are Odd, Error Correcting Codes, Distribution Classifier Defense, K-Winner Take All and Buffer Zones. Our investigation is done using two black-box adversarial models and six widely studied adversarial attacks for CIFAR-10 and Fashion-MNIST datasets. Our analyses show most recent defenses (7 out of 9) provide only marginal improvements in security (<25%), as compared to undefended networks. For every defense, we also show the relationship between the amount of data the adversary has at their disposal, and the effectiveness of adaptive black-box attacks. Overall, our results paint a clear picture: defenses need both thorough white-box and black-box analyses to be considered secure. We provide this large scale study and analyses to motivate the field to move towards the development of more robust black-box defenses.

Interest Groups and Elections

2019 ◽  
pp. 339-357
Author(s):  
Jeffrey M. Berry
Keyword(s):  
Political Parties ◽  
Interest Groups ◽  
Campaign Finance ◽  
Organizational Forms ◽  
Super Pacs ◽  
New Organizational Forms ◽  
Winner Take All ◽  
The Impact ◽  
Take All

The relationships between interest groups, political parties, and elections have always been dynamic, but in recent years change has accelerated in ways that have favored some interests over others. This chapter considers these developments as the result of a variety of factors, the most critical of which are the growth of polarization, a new legal landscape for campaign finance, and new organizational forms. The chapter goes on to suggest, that as bipartisanship has ebbed, elections have become winner-take-all affairs and interest groups are pushed to choose sides. The chapter further suggests that the rise of super PACs is especially notable as wealthy individuals have become increasingly important, single sources of campaign money, supplanting in part traditional interest groups, especially conventional PACs. It concludes that even as sums spent by super PACs and other interest groups have skyrocketed, the impact of their direct spending on persuading voters remains uncertain.

Class Politics, American-Style

2011 ◽  
Vol 9 (3) ◽  
pp. 659-662 ◽  
Author(s):  
Brian Waddell
Keyword(s):  
United States ◽  
Political Economy ◽  
Middle Class ◽  
Political Science ◽  
The United States ◽  
The Political ◽  
The Rich ◽  
State Of Affairs ◽  
Winner Take All ◽  
Take All

Jacob S. Hacker and Paul Pierson's Winner-Take-All Politics: How Washington Made the Rich Richer—And Turned Its Back on the Middle Class is both a work of political science and a contribution to broad public discussion of distributive politics. Its topic could not be more relevant to a US polity wracked by bitter partisan disagreements about taxes, social spending, financial regulation, social insecurity, and inequality. The political power of “the rich” is a theme of widespread public attention. The headline on the cover of the January–February 2011 issue of The American Interest—“Inequality and Democracy: Are Plutocrats Drowning Our Republic?”—is indicative. Francis Fukuyama's lead essay, entitled “Left Out,” clarifies that by “plutocracy,” the journal means “not just rule by the rich, but rule by and for the rich. We mean, in other words, a state of affairs in which the rich influence government in such a way as to protect and expand their own wealth and influence, often at the expense of others.” Fukuyama makes clear that he believes that this state of affairs obtains in the United States today.Readers of Perspectives on Politics will know that the topic has garnered increasing attention from political scientists in general and in our journal in particular. In March 2009, we featured a symposium on Larry Bartels's Unequal Democracy: The Political Economy of the New Gilded Age. And in December 2009, our lead article, by Jeffrey A. Winters and Benjamin I. Page, starkly posed the question “Oligarchy in the United States?” and answered it with an equally stark “yes.” Winner-Take-All Politics thus engages a broader scholarly discussion within US political science, at the same time that it both draws upon and echoes many “classic themes” of US political science from the work of Charles Beard and E. E. Schattschneider to Ted Lowi and Charles Lindblom.In this symposium, we have brought together a group of important scholars and commentators who offer a range of perspectives on the book and on the broader themes it engages. While most of our discussants are specialists on “American politics,” we have also sought out scholars beyond this subfield. Our charge to the discussants is to evaluate the book's central claims and evidence, with a focus on three related questions: 1) How compelling is its analysis of the “how” and “why” of recent US public policy and its “turn” in favor of “the rich” and against “the middle class”? 2) How compelling is its critique of the subfield of “American politics” for its focus on the voter–politician linkage and on “politics as spectacle” at the expense of an analysis of “politics as organized combat”? 3) And do you agree with its argument that recent changes in US politics necessitate a different, more comparative, and more political economy–centered approach to the study of US politics?—Jeffrey C. Isaac, Editor

Sign in / Sign up

Export Citation Format

Share Document