Learn More
We propose a new framework for estimating generative models via an adversar-ial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the(More)
MXNet is a multi-language machine learning (ML) library to ease the development of ML algorithms, especially for deep neural networks. Embedded in the host language, it blends declarative symbolic expression with imperative tensor computation. It offers auto differentiation to derive gradients. MXNet is computation and memory efficient and runs on various(More)
In this paper we investigate the performance of different types of rectified activation functions in convolutional neural network: standard rectified linear unit (ReLU), leaky rectified linear unit (Leaky ReLU), parametric rectified linear unit (PReLU) and a new ran-domized leaky rectified linear units (RReLU). We evaluate these activation function on(More)
We propose a new framework for estimating generative models via an adversar-ial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the(More)
The ICML 2013 Workshop on Challenges in Representation Learning(1) focused on three challenges: the black box learning challenge, the facial expression recognition challenge, and the multimodal learning challenge. We describe the datasets created for these challenges and summarize the results of the competitions. We provide suggestions for organizers of(More)
The aim of the present study was to compare the changes in the levels of 27 aqueous humor cytokines between diabetic patients with macular edema (ME) and diabetic patients without ME. Undiluted aqueous humor samples were obtained from 68 consecutive type 2 diabetic patients without ME and 56 consecutive type 2 diabetic patients with ME. The concentrations(More)
BACKGROUND Since late 2003, highly pathogenic avian influenza (HPAI) outbreaks caused by infection with H5N1 virus has led to the deaths of millions of poultry and more than 10 thousands of wild birds, and as of 18-March 2008, at least 373 laboratory-confirmed human infections with 236 fatalities, have occurred. The unrestrained worldwide spread of this(More)
Plant growth requires cell wall extension. The cotton AtRD22-Like 1 gene GhRDL1, predominately expressed in elongating fiber cells, encodes a BURP domain-containing protein. Here, we show that GhRDL1 is localized in cell wall and interacts with GhEXPA1, an α-expansin functioning in wall loosening. Transgenic cotton overexpressing GhRDL1 showed an increase(More)
In this paper, we propose a method, learning with adversary, to learn a robust network. Our method takes finding adversarial examples as its mediate step. A new and simple way of finding adversarial examples are presented and experimentally shown to be more 'efficient'. Lastly, experimental results shows our learning method greatly improves the robustness(More)