In spite of remarkable success of the convolutional neural networks on semantic segmentation, they suffer from catastrophic forgetting: a significant performance drop for the already learned classes when new classes are added on the data, having no annotations for the old classes. We propose an incremental learning methodology, enabling to learn segmenting new classes without hindering dense labeling abilities for the previous classes, although the entire previous data are not accessible. The key points of the proposed approach are adapting the network to learn new as well as old classes on the new training data, and allowing it to remember the previously learned information for the old classes. For adaptation, we keep a frozen copy of the previously trained network, which is used as a memory for the updated network in absence of annotations for the former classes. The updated network minimizes a loss function, which balances the discrepancy between outputs for the previous classes from the memory and updated networks, and the mis-classification rate between outputs for the new classes from the updated network and the new ground-truth. For remembering, we either regularly feed samples from the stored, little fraction of the previous data or use the memory network, depending on whether the new data are collected from completely different geographic areas or from the same city. Our experimental results prove that it is possible to add new classes to the network, while maintaining its performance for the previous classes, despite the whole previous training data are not available.

The world is witnessing an unprecedented growth of cyber-physical systems (CPS), which are foreseen to revolutionize our world {via} creating new services and applications in a variety of sectors such as environmental monitoring, mobile-health systems, intelligent transportation systems and so on. The {information and communication technology }(ICT) sector is experiencing a significant growth in { data} traffic, driven by the widespread usage of smartphones, tablets and video streaming, along with the significant growth of sensors deployments that are anticipated in the near future. {It} is expected to outstandingly increase the growth rate of raw sensed data. In this paper, we present the CPS taxonomy {via} providing a broad overview of data collection, storage, access, processing and analysis. Compared with other survey papers, this is the first panoramic survey on big data for CPS, where our objective is to provide a panoramic summary of different CPS aspects. Furthermore, CPS {require} cybersecurity to protect {them} against malicious attacks and unauthorized intrusion, which {become} a challenge with the enormous amount of data that is continuously being generated in the network. {Thus, we also} provide an overview of the different security solutions proposed for CPS big data storage, access and analytics. We also discuss big data meeting green challenges in the contexts of CPS.

Recently, several logit regularization methods have been proposed in [Kannan et al., 2018] to improve the adversarial robustness of classifiers. We show that the proposed computationally fast methods - Clean Logit Pairing (CLP) and Logit Squeezing (LSQ) - just make the gradient-based optimization problem of crafting adversarial examples harder, without providing actual robustness. For Adversarial Logit Pairing (ALP) we find that it can give indeed robustness against adversarial examples and we study it in different settings. Especially, we show that ALP may provide additional robustness when combined with adversarial training. However, the increase is much smaller than claimed by [Kannan et al., 2018]. Finally, our results suggest that evaluation against an iterative PGD attack relies heavily on the parameters used and may result in false conclusions regarding the robustness.

The problem addressed in this paper is how to calculate the amount of personnel required to ensure the activity of a home health care (HHC) center on a tactical horizon. Design of quantitative approaches for this question is challenging. The number of caregivers has to be determined for each profession in order to balance the coverage of patients in a region and the workforce cost over several months. Unknown demand in care and spatial dimensions, combination of skills to cover a care and individual trips visiting patients make the underlaying optimization problem very hard. Few studies are dedicated to staff dimensioning for HHC compared to patient to nurses assignment/sequencing and centers location problems. We propose an original two-stage approach based on integer linear stochastic programming, that exploits historical medical data. The first stage calculates (near-)optimal levels of resources for possible demand scenarios , while the second stage computes the optimal number of caregiver for each profession to meet a target coverage indicator. For decision-makers, our algorithm gives the number of employees for each category required to satisfy the demand without any recourse (overtime, external resources) with fixed probability and confidence interval. The approach has been tested on various instances built from data of the French agency of hospitalization data (ATIH).

Daniel Kahneman was awarded the Nobel Prize in Economic Science in 2002 for his pioneering studies on the identification and estimation of the psychological factors that influence human behaviour under uncertainty, which led to the birth of a new domain called behavioural economics. Cognitive psychologists have assumed for years, often implicitly, that complex cognitive processes, like human judgement and decision-making (DM), have to be modelled by combining set-theoretic structures and should obey to mathematical relations that resemble those typically used in logic, formalized by Boole (Boolean logic), and probability, axiomatized by Kolmogorov (Kolmogorovian probability) [1]. These structures are known in physics as classical structures: they were originally used in classical physics, and later extended to statistics, psychology, economics, finance and computer science. Classical structures are also implicitly assumed in the so-called Bayesian approach, according to which any source of uncertainty can be formalized probabilistically, while people update knowledge according to the Bayes law of Kolmogorovian probability. Finally, classical structures are the building blocks of subjective expected utility theory (SEUT), providing both the descriptive and the normative foundations of rational DM: in situations of uncertainty, people (should) choose as if they maximized EU with respect to a unique probability measure, satisfying the axioms of Kolmogorov and interpreted as their subjective probability [2, 3]. However, on the one side, empirical research in cognitive psychology has revealed that classical structures are not generally able to model human judgements and decisions, thus making problematical the 1 interpretation of a wide range of cognitive phenomena in terms of standard logic and probability theory. On the other side, Kahneman, Tversky and other authors suggested that these empirical deviations from classicality are "true errors" of human reasoning, whence the use of terms like "effect", "fallacy", "paradox", "contradiction", etc., to refer to such phenomena [4, 5].

Efficient exploration is an unsolved problem in Reinforcement Learning. We introduce Model-Based Active eXploration (MAX), an algorithm that actively explores the environment. It minimizes data required to comprehensively model the environment by planning to observe novel events, instead of merely reacting to novelty encountered by chance. Non-stationarity induced by traditional exploration bonus techniques is avoided by constructing fresh exploration policies only at time of action. In semi-random toy environments where directed exploration is critical to make progress, our algorithm is at least an order of magnitude more efficient than strong baselines.

It's the stuff of horror films: an intruder in your house, impossible to find but undeniably somewhere, watching you at your most private moments. Or perhaps it's the plot of a thriller, where you are recruited into international crime without even knowing it, at the behest of smart criminals. If the worst fears about the prevalence of weakly secured smart home gadgets materialise, those terrifying situations could become all too real. As we fill our homes with internet-enabled and smart devices, we are opening ourselves up to attacks that exploit houses themselves – and we might not even realise they are happening. Everything from washing machines to baby monitors is being hooked up to the internet by companies convinced that features such as remote control and artificial intelligence will make our lives easier and safer.

In the future AI could develop a will of its own, a will that is in conflict with ours. These were some of the final words from the late, great Stephen Hawking that were published earlier this month when he warned about the consequences of unregulated artificial intelligence. Hawking also predicted the rise of the'superhuman' who will initially rely on AI and genetics and then eventually escape earth; hopefully to do a better job on a new planet than we have done to date on ours. It will be some time before superhuman are with us, but AI certainly is and it's completely transforming the world and how humans operate within it. This is not a episode of Black Mirror, these are the times we live in.

Our mission is to provide a novel artistic painting tool that allows everyone to create and share artistic pictures with just a few clicks. We are five researchers working at the interface of neuroscience and artificial intelligence, based at the University of Tübingen (Germany), École polytechnique fédérale de Lausanne (Switzerland) and Université catholique de Louvain (Belgium).

Apple CEO, Tim Cook spoke up for privacy at a conference of European privacy commissioners in Brussels this morning. The themes of this year's conference is "Debating Ethics: Dignity and Respect in Data Driven Life", Cook is the first tech CEO to serve as the keynote speaker for the conference and was invited to speak. He talked about data, put in a bid for a bill of U.S. digital rights, slammed competitors for profiting while unleashing powerfully negative forces, and spoke up for a GDPR-style privacy protection in the U.S. What follows is the transcript of his speech. "It is an honor to be here with you today in this grand hall…a room that represents what is possible when people of different backgrounds, histories, and philosophies come together to build something bigger than themselves. "I am deeply grateful to our hosts.