University of Houston

About: University of Houston is a education organization based out in Houston, Texas, United States. It is known for research contribution in the topics: Population & Poison control. The organization has 23074 authors who have published 53903 publications receiving 1641968 citations.

Semi-Supervised Deep Learning Using Pseudo Labels for Hyperspectral Image Classification

Abstract: Deep learning has gained popularity in a variety of computer vision tasks. Recently, it has also been successfully applied for hyperspectral image classification tasks. Training deep neural networks, such as a convolutional neural network for classification requires a large number of labeled samples. However, in remote sensing applications, we usually only have a small amount of labeled data for training because they are expensive to collect, although we still have abundant unlabeled data. In this paper, we propose semi-supervised deep learning for hyperspectral image classification—our approach uses limited labeled data and abundant unlabeled data to train a deep neural network. More specifically, we use deep convolutional recurrent neural networks (CRNN) for hyperspectral image classification by treating each hyperspectral pixel as a spectral sequence. In the proposed semi-supervised learning framework, the abundant unlabeled data are utilized with their pseudo labels (cluster labels). We propose to use all the training data together with their pseudo labels to pre-train a deep CRNN, and then fine-tune using the limited available labeled data. Further, to utilize spatial information in the hyperspectral images, we propose a constrained Dirichlet process mixture model (C-DPMM), a non-parametric Bayesian clustering algorithm, for semi-supervised clustering which includes pairwise must-link and cannot-link constraints—this produces high-quality pseudo-labels, resulting in improved initialization of the deep neural network. We also derived a variational inference model for the C-DPMM for efficient inference. Experimental results with real hyperspectral image data sets demonstrate that the proposed semi-supervised method outperforms state-of-the-art supervised and semi-supervised learning methods for hyperspectral classification.

Out-of-Sample Exchange Rate Predictability with Taylor Rule Fundamentals

Abstract: An extensive literature that studied the performance of empirical exchange rate models following Meese and Rogoff's (1983a) seminal paper has not convincingly found evidence of out-of-sample exchange rate predictability. This paper extends the conventional set of models of exchange rate determination by investigating predictability of models that incorporate Taylor rule fundamentals. We find evidence of short term predictability for 11 out of 12 currencies vis-a-vis the U.S. dollar over the post-Bretton Woods float, with the strongest evidence coming from specifications that incorporate heterogeneous coefficients and interest rate smoothing. The evidence of predictability is much stronger with Taylor rule models than with conventional interest rate, purchasing power parity, or monetary models.

Rural Windfall or a New Resource Curse? Coca, Income, and Civil Conflict in Colombia

Abstract: Natural and agricultural resources for which there is a substantial black market, such as coca, opium, and diamonds, appear especially likely to be exploited by the parties to a civil conflict. Even legally traded commodities such as oil and timber have been linked to civil war. On the other hand, these resources may also provide one of the few reliable sources of income in the countryside. In this paper, we study the economic and social consequences of a major exogenous shift in the production of one such resource - coca paste - into Colombia, where most coca leaf is now harvested. Our analysis shows that this shift generated only modest economic gains in rural areas, primarily in the form of increased self-employment earnings and increased labor supply by teenage boys. The results also suggest that the rural areas which saw accelerated coca production subsequently became more violent, while urban areas were affected little. The acceleration in violence is greater in departments (provinces) where there was a pre-coca guerilla presence. Taken together, these findings are consistent with the view that the Colombian civil conflict is fueled by the financial opportunities that coca provides, and that the consequent rent-seeking activity by combatants limits the economic gains from coca cultivation.