Research Interests:
Research Interests:
Research Interests:
Provisioning the city of Rome required complex systems of management and procurement. Despite the importance of wood for baking the grain of the annona and heating the water for baths (among other applications), there is little evidence... more
Provisioning the city of Rome required complex systems of management and procurement. Despite the importance of wood for baking the grain of the annona and heating the water for baths (among other applications), there is little evidence for the human or ecological connections that supplied Rome with this crucial fuel. By examining parallel institutions, this article provides a model for explaining how the city of Rome met the energy demands of its massive population. We suggest some ways urban labour was linked to Italian forests, creating a sustainable mode of supply that lasted for centuries.
Research Interests:
Research Interests: Law and Applied Economics
Research Interests:
Computationally, we have developed a sparse CNN implementation that make it practical to train CNNs with many layers of max-pooling. Extending the MNIST dataset by translations, our sparse CNN gets a test error of 0.31%.
Dropout is a popular technique for regularizing artificial neural networks. Dropout networks are generally trained by minibatch gradient descent with a dropout mask turning off some of the units---a different pattern of dropout is applied... more
Dropout is a popular technique for regularizing artificial neural networks. Dropout networks are generally trained by minibatch gradient descent with a dropout mask turning off some of the units---a different pattern of dropout is applied to every sample in the minibatch. We explore a very simple alternative to the dropout mask. Instead of masking dropped out units by setting them to zero, we perform matrix multiplication using a submatrix of the weight matrix---unneeded hidden units are never calculated. Performing dropout batchwise, so that one pattern of dropout is used for each sample in a minibatch, we can substantially reduce training times. Batchwise dropout can be used with fully-connected and convolutional neural networks.
Research Interests:
The Intelligent Investor © 1949 by Benjamin Graham; Foreword © 2005 by John C. Bogle. All rights reserved. Printed in the United States of America. No part of this book may be used or re-produced in any manner whatsoever without written... more
The Intelligent Investor © 1949 by Benjamin Graham; Foreword © 2005 by John C. Bogle. All rights reserved. Printed in the United States of America. No part of this book may be used or re-produced in any manner whatsoever without written permission except in the case of ...
... and the Caucasus Research Resources Center for in-‐ country support, and to the wonderful Capital and Conflict: Georgia team: Maya Baramidze, Amiran Chanchibadze, Anna Gogokhia, David Kokiashvili, Alexandre Kukhianidze, Anastasia... more
... and the Caucasus Research Resources Center for in-‐ country support, and to the wonderful Capital and Conflict: Georgia team: Maya Baramidze, Amiran Chanchibadze, Anna Gogokhia, David Kokiashvili, Alexandre Kukhianidze, Anastasia Laitadze, Giorgi Mekerishvili, Maia ...
Research Interests:
Convolutional networks almost always incorporate some form of spatial pooling, and very often it is alpha times alpha max-pooling with alpha=2. Max-pooling act on the hidden layers of the network, reducing their size by an integer... more
Convolutional networks almost always incorporate some form of spatial pooling, and very often it is alpha times alpha max-pooling with alpha=2. Max-pooling act on the hidden layers of the network, reducing their size by an integer multiplicative factor alpha. The amazing by-product of discarding 75% of your data is that you build into the network a degree of invariance with respect to translations and elastic distortions. However, if you simply alternate convolutional layers with max-pooling layers, performance is limited due to the rapid reduction in spatial size, and the disjoint nature of the pooling regions. We have formulated a fractional version of max-pooling where alpha is allowed to take non-integer values. Our version of max-pooling is stochastic as there are lots of different ways of constructing suitable pooling regions. We find that our form of fractional max-pooling reduces overfitting on a variety of datasets: for instance, we improve on the state-of-the art for CIFAR-100 without even using dropout.
Research Interests:
Research Interests:
Research Interests:
Studies of regime type and war show that democracies tend to win the wars they fight, but questions remain about why this is the case. A simple, if underappreciated, explanation for democratic success is that democracies fight alongside... more
Studies of regime type and war show that democracies tend to win the wars they fight, but questions remain about why this is the case. A simple, if underappreciated, explanation for democratic success is that democracies fight alongside larger and more powerful coalitions. Coalition partners bring additional material capabilities and may also provide intangible benefits to the war effort, such as increased legitimacy or confidence. Democracies may also find it less costly to join coalitions, as democratic war aims may be easier to apportion among the victors without diluting the spoils. Evaluating our hypotheses in a sample of all wars (or all militarized disputes) during the period 1816–2000, we find that democracies fight alongside larger coalitions and that states fighting alongside larger coalitions are more likely to win major contests. Coalition size subsumes most (and in some specifications all) of the direct effect of regime type on victory.
Research Interests:
Research Interests:
Research Interests:
... but also the actual costs paid by firms who wish to invest without shouldering the burden of political risk themselves.8 ONDD data are used in a similar context by Jensen (2008) and Jensen and Young (2008). ONDD scores ...
Research Interests:
Research Interests: Economics and Social Work
Research Interests:
Bondage-discipline/Dominance-submission/sadomasochism (BDSM) is an often misunderstood and misrepresented social phenomenon warranting further discourse and study. Community-based research that engages member perspective can assist in... more
Bondage-discipline/Dominance-submission/sadomasochism (BDSM) is an often misunderstood and misrepresented social phenomenon warranting further discourse and study. Community-based research that engages member perspective can assist in understanding socially marginalized experiences. The current study examined the role, meaning, and function of BDSM communities from the perspective of self-identified members. Seven nominal group technique workshops were conducted representing a variety of practitioner experiences and identities. Workshops involved 48 participants and resulted in the generation of 133 unique terms describing the role of BDSM communities in their lives. Terms were coded using a five-step procedure involving both academic and community members. A total of 15 categories were identified and included domains such as acceptance, sexual expression, friendship, safety, and sharing of educational knowledge. Results underscore the multifaceted nature of the role of such communi...
Research Interests:
Research Interests:
ABSTRACT Abstract will be provided by author.
Research Interests: Investment and FDI
ABSTRACT
Research Interests:
The number of balls in the ith box is bi . We will write Bi = Bi(t) for the corre- sponding random variable. We will call this the mean-field zero-range process (MFZRP). The zero-range process is normally defined on a directed graph, with... more
The number of balls in the ith box is bi . We will write Bi = Bi(t) for the corre- sponding random variable. We will call this the mean-field zero-range process (MFZRP). The zero-range process is normally defined on a directed graph, with balls jumping along the edges. In contrast, our process is implicitly defined on a complete graph—balls can
Research Interests:
ABSTRACT
Research Interests:
The first-hand experience of operating people combined with well prepared charts of their processes can take the guesswork out of process improvement. Process charts provide cross-functional teams of operating people with a comprehensive... more
The first-hand experience of operating people combined with well prepared charts of their processes can take the guesswork out of process improvement. Process charts provide cross-functional teams of operating people with a comprehensive picture enabling them to view their work from the perspective of the organization rather than from their particular work areas. At the same time the charts break
Research Interests:
Research Interests:
ABSTRACT