Generalized Data Representation (GDR) google
Deep learning (DL) based autoencoder is a promising architecture to implement end-to-end communication systems. In this paper, we focus on the fundamental problems of DL-based communication systems, including high rate transmission and performance analysis. To address the limited data rate issue, we first consider the error rate constraint and design a transmission algorithm to adaptively select the transmission vectors to maximize the data rate for various channel scenarios. Furthermore, a novel generalized data representation (GDR) scheme is proposed to improve the data rate of DL-based communication systems. Then, we analyze the effect of signal-to-noise ratio (SNR) and mean squared error performance of the proposed DL-based communication systems. Finally, numerical results show that the proposed adaptive transmission and GDR schemes achieve higher data rate and have lower training complexity than the conventional one-hot vector scheme. Both the new schemes and the conventional scheme have comparable block error rate (BLER) performance. According to both theoretical analysis and simulated results, it is suggested that low or wide-range training SNR is beneficial to attain good BLER performance for practical transmission with various channel scenarios. …

PathGAN google
We introduce PathGAN, a deep neural network for visual scanpath prediction trained on adversarial examples. A visual scanpath is defined as the sequence of fixation points over an image defined by a human observer with its gaze. PathGAN is composed of two parts, the generator and the discriminator. Both parts extract features from images using off-the-shelf networks, and train recurrent layers to generate or discriminate scanpaths accordingly. In scanpath prediction, the stochastic nature of the data makes it very difficult to generate realistic predictions using supervised learning strategies, but we adopt adversarial training as a suitable alternative. Our experiments prove how PathGAN improves the state of the art of visual scanpath prediction on the iSUN and Salient360! datasets. Source code and models are available at https://…/pathgan

Exposure google
Machine learning models based on neural networks and deep learning are being rapidly adopted for many purposes. What those models learn, and what they may share, is a significant concern when the training data may contain secrets and the models are public — e.g., when a model helps users compose text messages using models trained on all users’ messages. This paper presents exposure: a simple-to-compute metric that can be applied to any deep learning model for measuring the memorization of secrets. Using this metric, we show how to extract those secrets efficiently using black-box API access. Further, we show that unintended memorization occurs early, is not due to over-fitting, and is a persistent issue across different types of models, hyperparameters, and training strategies. We experiment with both real-world models (e.g., a state-of-the-art translation model) and datasets (e.g., the Enron email dataset, which contains users’ credit card numbers) to demonstrate both the utility of measuring exposure and the ability to extract secrets. Finally, we consider many defenses, finding some ineffective (like regularization), and others to lack guarantees. However, by instantiating our own differentially-private recurrent model, we validate that by appropriately investing in the use of state-of-the-art techniques, the problem can be resolved, with high utility. …