The European Union's executive body is doubling down on its push for platforms to pre-filter the Internet, publishing a proposal today for all websites to monitor uploads in order to be able to quickly remove terrorist uploads. The Commission handed platforms an informal one-hour rule for removing terrorist content back in March. It's now proposing turning that into a law to prevent such content spreading its violent propaganda over the Internet. For now the'rule of thumb' regime continues to apply. But it's putting meat on the bones of its thinking, fleshing out a more expansive proposal for a regulation aimed at "preventing the dissemination of terrorist content online".

Video Friday is your weekly selection of awesome robotics videos, collected by your Automaton bloggers. We'll also be posting a weekly calendar of upcoming robotics events for the next few months; here's what we have so far (send us your events!): Let us know if you have suggestions for next week, and enjoy today's videos. Maurice Fallon and Ioannis Havoutis from the Dynamic Robot Systems Group at the University of Oxford have a shiny new ANYmal that they've been putting to work doing useful stuff: I know Boston Dynamics gets all the attention, but in terms of practical quadrupeds, ANYmal is getting a lot of solid work done. Might want to teach it not to impale itself through the neck with traffic cones, though.

Bender is one of the robot arms that a team of Stanford researchers is using to test two frameworks that, together, could make it faster and easier to teach robots basic skills. The RoboTurk framework allows people to direct the robot arms in real time with a smartphone and a browser by showing the robot how to carry out tasks like picking up objects. SURREAL speeds the learning process by running multiple experiences at once, essentially allowing the robots to learn from many experiences simultaneously. "With RoboTurk and SURREAL, we can push the boundary of what robots can do by combining lots of data collected by humans and coupling that with large-scale reinforcement learning," said Mandlekar, a member of the team that developed the frameworks. The group will be presenting RoboTurk and SURREAL Oct. 29 at the conference on robot learning in Zurich, Switzerland.

On January 17 1803, a young man named George Forster was hanged for murder at Newgate prison in London. After his execution, as often happened, his body was carried ceremoniously across the city to the Royal College of Surgeons, where it would be publicly dissected. What actually happened was rather more shocking than simple dissection though - Forster was going to be electrified. Giovanni Aldini's experiments with a human corpse - one of the first, and most controversial, to attempt to reanimate a human using electricity Luigi Galvani was an Italian physician who demonstrated what we now understand to be the electrical basis of nerve impulses, when he made frog muscles twitch by jolting them with a spark from an electrostatic machine. Galvani one day observed his assistant using a scalpel on a nerve in a frog's leg; when a nearby electric generator created a spark, the frog's leg twitched, prompting Galvani to develop his famous experiment.

Deshpande et al. presented a k(ln R + 1) approximation bound for Stochastic Submodular Cover, where k is the state set size, R is the maximum utility of a single item, and the utility function is integer-valued. This bound is similar to the ln Q/(eta+1) bound given by Golovin and Krause, whose analysis was recently found to have an error. Here Q >= R is the goal utility and eta is the minimum gap between Q and any attainable utility Q' < Q. We revisit the proof of the k(ln R + 1) bound of Deshpande et al., fill in the details of the proof of a key lemma, and prove two bounds for real-valued utility functions: k(ln R_1 + 1) and (ln R_E + 1). Here R_1 equals the maximum ratio between the largest increase in utility attainable from a single item, and the smallest non-zero increase attainable from that same item (in the same state). The quantity R_E equals the maximum ratio between the largest expected increase in utility from a single item, and the smallest non-zero expected increase in utility from that same item. Our bounds apply only to the stochastic setting with independent states.

Generative adversarial network (GAN) is a minimax game between a generator mimicking the true model and a discriminator distinguishing the samples produced by the generator from the real training samples. Given an unconstrained discriminator able to approximate any function, this game reduces to finding the generative model minimizing a divergence measure, e.g. the Jensen-Shannon (JS) divergence, to the data distribution. However, in practice the discriminator is constrained to be in a smaller class $\mathcal{F}$ such as neural nets. Then, a natural question is how the divergence minimization interpretation changes as we constrain $\mathcal{F}$. In this work, we address this question by developing a convex duality framework for analyzing GANs. For a convex set $\mathcal{F}$, this duality framework interprets the original GAN formulation as finding the generative model with minimum JS-divergence to the distributions penalized to match the moments of the data distribution, with the moments specified by the discriminators in $\mathcal{F}$. We show that this interpretation more generally holds for f-GAN and Wasserstein GAN. As a byproduct, we apply the duality framework to a hybrid of f-divergence and Wasserstein distance. Unlike the f-divergence, we prove that the proposed hybrid divergence changes continuously with the generative model, which suggests regularizing the discriminator's Lipschitz constant in f-GAN and vanilla GAN. We numerically evaluate the power of the suggested regularization schemes for improving GAN's training performance.

We examine the relationship between the energy landscape of neural networks and their robustness to adversarial attacks. Combining energy landscape techniques developed in computational chemistry with tools drawn from formal methods, we produce empirical evidence that networks corresponding to lower-lying minima in the landscape tend to be more robust. The robustness measure used is the inverse of the sensitivity measure, which we define as the volume of an over-approximation of the reachable set of network outputs under all additive $l_{\infty}$ bounded perturbations on the input data. We present a novel loss function which contains a weighted sensitivity component in addition to the traditional task-oriented and regularization terms. In our experiments on standard machine learning and computer vision datasets (e.g., Iris and MNIST), we show that the proposed loss function leads to networks which reliably optimize the robustness measure as well as other related metrics of adversarial robustness without significant degradation in the classification error.

Activation functions influence behavior and performance of DNNs. Nonlinear activation functions, like Rectified Linear Units (ReLU), Exponential Linear Units (ELU) and Scaled Exponential Linear Units (SELU), outperform the linear counterparts. However, selecting an appropriate activation function is a challenging problem, as it affects the accuracy and the complexity of the given DNN. In this paper, we propose a novel methodology to automatically select the best-possible activation function for each layer of a given DNN, such that the overall DNN accuracy, compared to considering only one type of activation function for the whole DNN, is improved. However, an associated scientific challenge in exploring all the different configurations of activation functions would be time and resource-consuming. Towards this, our methodology identifies the Evaluation Points during learning to evaluate the accuracy in an intermediate step of training and to perform early termination by checking the accuracy gradient of the learning curve. This helps in significantly reducing the exploration time during training. Moreover, our methodology selects, for each layer, the dropout rate that optimizes the accuracy. Experiments show that we are able to achieve on average 7% to 15% Relative Error Reduction on MNIST, CIFAR-10 and CIFAR-100 benchmarks, with limited performance and power penalty on GPUs.

Designing and modifying complex hull forms for optimal vessel performances have been a major challenge for naval architects. In the present study, Principal Component Analysis (PCA) is introduced to compress the geometric representation of a group of existing vessels, and the resulting principal scores are manipulated to generate a large number of derived hull forms, which are evaluated computationally for their calm-water performances. The results are subsequently used to train a Deep Neural Network (DNN) to accurately establish the relation between different hull forms and their associated performances. Then, based on the fast, parallel DNN-based hull-form evaluation, the large-scale search for optimal hull forms is performed.

Tree-based ensemble methods, as Random Forests and Gradient Boosted Trees, have been successfully used for regression in many applications and research studies. Furthermore, these methods have been extended in order to deal with uncertainty in the output variable, using for example a quantile loss in Random Forests (Meinshausen, 2006). To the best of our knowledge, no extension has been provided yet for dealing with uncertainties in the input variables, even though such uncertainties are common in practical situations. We propose here such an extension by showing how standard regression trees optimizing a quadratic loss can be adapted and learned while taking into account the uncertainties in the input. By doing so, one no longer assumes that an observation lies into a single region of the regression tree, but rather that it belongs to each region with a certain probability. Experiments conducted on several data sets illustrate the good behavior of the proposed extension.