The widespread application of deep learning has changed the landscape of computation in the data center. In particular, personalized recommendation for content ranking is now largely accomplished leveraging deep neural networks. However, despite the importance of these models and the amount of compute cycles they consume, relatively little research attention has been devoted to systems for recommendation. To facilitate research and to advance the understanding of these workloads, this paper presents a set of real-world, productionscale DNNs for personalized recommendation coupled with relevant performance metrics for evaluation. In addition to releasing a set of open-source workloads, we conduct indepth analysis that underpins future system design and optimization for at-scale recommendation: Inference latency varies by 60% across three Intel server generations, batching and co-location of inferences can drastically improve latency-bounded throughput, and the diverse composition of recommendation models leads to different optimization strategies.Preprint. Under submission.

No abstract

We assessed the effects of wind conditions on stopover decisions and fuel stores of migratory shorebirds at Chongming Dongtan in the south Yellow Sea along the East Asian-Australasian Flyway. In spring and autumn, wind directions differed among altitudes and wind speed generally increased with altitude. Numbers of shorebirds were related to wind effects at low altitudes (on the ground and at 300 and 800 m above the ground), wind effects at 300 m being the best predictor of shorebird numbers. In spring, total number of shorebirds and numbers of the four most abundant shorebird species were negatively related to wind assistance at low altitudes, more birds departing when tailwinds prevailed and more arriving when headwinds prevailed. In autumn, however, total number of shorebirds and numbers of the four most abundant species were positively related to wind assistance at low altitudes, more birds departing and more arriving with tailwinds than with headwinds. When tailwinds prevailed, the number of arriving birds was higher than the number of departing birds. The fuel stores of captured shorebirds, represented by their body mass, was related to wind effects and change in wind conditions between two consecutive days in both spring and autumn, captured birds being heavier when headwinds prevailed than in tailwind conditions, and when the wind conditions became less favourable for flight between two consecutive days. Our results suggest that wind conditions affect stopover decisions and fuel stores, and thus the optimal migration and fuel deposition strategies of migratory shorebirds.

Recent advances in machine learning, especially techniques such as deep neural networks, are enabling a range of emerging applications. One such example is autonomous driving, which often relies on deep learning for perception. However, deep learning-based perception has been shown to be vulnerable to a host of subtle adversarial manipulations of images. Nevertheless, the vast majority of such demonstrations focus on perception that is disembodied from end-to-end control. We present novel endto-end attacks on autonomous driving in simulation, using simple physically realizable attacks: the painting of black lines on the road. These attacks target deep neural network models for endto-end autonomous driving control. A systematic investigation shows that such attacks are easy to engineer, and we describe scenarios (e.g., right turns) in which they are highly effective. We define several objective functions that quantify the success of an attack and develop techniques based on Bayesian Optimization to efficiently traverse the search space of higher dimensional attacks. Additionally, we define a novel class of hijacking attacks, where painted lines on the road cause the driverless car to follow a target path. Through the use of network deconvolution, we provide insights into the successful attacks, which appear to work by mimicking activations of entirely different scenarios. Our code is available on https://github.com/xz-group/AdverseDrive Index Terms-machine learning, adversarial examples, autonomous driving, end-to-end learning, bayesian optimization