Clementina works as a UX designer for Proximus, a Belgian telco provider. In the past she was a consultant in the field of UX and Service design and she has been working for clients like Sony, the European Commission and the Flemish Government. She has been a speaker in various international conferences such as EuroIA2016, IA Summit Rome 2016, WIAD Zurich 2017, WIAD Bari 2018.

The Army offered a $100,000 prize for a solution to an increasingly tough problem for commanders in the field: In a battlefield dense with electromagnetic signals, is there a better way to distinguish friendly transmissions from hostile attacks? There is, according to a team of eight engineers from Aerospace Corporation, based in El Segundo, Calif. They won the prize by correctly detecting and classifying the greatest number of radio frequency signals using a combination of signal processing and artificial intelligence algorithms. The competition, known as the "Blind Signal Classification Challenge," was sponsored by the Army's Rapid Capabilities Office, a small organization that looks for ways to apply commercial technology to solve military problems. When the challenge kicked off in April, the Army gave all 49 competitors a large amount of recordings of various types of radio signals to use as "training data" so they could develop their algorithms.

We are in the midst of a significant shift in pharmaceutical drug development, with many leading companies focusing increasingly on orphan drugs targeted at small niche markets. An orphan disease is a medical condition or disorder that affects less than 200,000 people in the US. The National Institutes of Health (NIH) has classified as many as 7,000 medical conditions as orphan diseases. Although a rare disease affects only a small population, the collective number of rare diseases affects as many as 25 million people. This mammoth-sized number of people requiring niche drugs is a serious public health concern.

When you return to school after summer break, it may feel like you forgot everything you learned the year before. But if you learned like an AI system does, you actually would have -- as you sat down for your first day of class, your brain would take that as a cue to wipe the slate clean and start from scratch. AI systems' tendency to forget the things it previously learned upon taking on new information is called catastrophic forgetting. See, cutting-edge algorithms learn, so to speak, after analyzing countless examples of what they're expected to do. A facial recognition AI system, for instance, will analyze thousands of photos of people's faces, likely photos that have been manually annotated, so that it will be able to detect a face when it pops up in a video feed.

HEIDELBERG, Germany (Reuters) - Armed with a mouse and computer screen instead of a scalpel and operating theater, cardiologist Benjamin Meder carefully places the electrodes of a pacemaker in a beating, digital heart. Using this "digital twin" that mimics the electrical and physical properties of the cells in patient 7497's heart, Meder runs simulations to see if the pacemaker can keep the congestive heart failure sufferer alive - before he has inserted a knife. The digital heart twin developed by Siemens Healthineers is one example of how medical device makers are using artificial intelligence (AI) to help doctors make more precise diagnoses as medicine enters an increasingly personalized age. The challenge for Siemens Healthineers and rivals such as Philips and GE Healthcare is to keep an edge over tech giants from Alphabet's Google to Alibaba that hope to use big data to grab a slice of healthcare spending. With healthcare budgets under increasing pressure, AI tools such as the digital heart twin could save tens of thousands of dollars by predicting outcomes and avoiding unnecessary surgery.

"Robot war on cancer" is the headline on the front of Saturday's Daily Express. The paper reports that scientists from the Institute of Cancer Research in London have designed a new computer tool that can learn to predict how tumours will grow, evolve and spread. Dr Andrea Sottoriva, who led the research, likens their work to a game of chess, saying: "The best chance we have of beating cancer is to predict the next move." The i says the breakthrough will transform care for millions of people and boost survival chances. The Sun's main story celebrates new baby joy for the parents of Alfie Evans - the toddler who was at the centre of a legal battle over his care before he died of a degenerative brain condition.

Scientists have developed an artificial intelligence system that can allow robots to interact with autistic children undergoing therapy. People with autism see, hear and feel the world differently from other people, which affects how they interact with others. This makes communication-centred activities quite challenging for children with autism spectrum conditions (ASCs). To address this challenge, therapists recently began to use humanoid robots in therapy sessions. However, existing robots lack the ability to autonomously engage with children, which is vital for improving the therapy.

Armed with a mouse and computer screen instead of a scalpel and operating theatre, cardiologist Benjamin Meder carefully places the electrodes of a pacemaker in a beating, digital heart. Using this'digital twin' that mimics the electrical and physical properties of the cells in patient 7497's heart, Meder runs simulations to see if the pacemaker can keep the congestive heart failure sufferer alive - before he has inserted a knife. The digital heart twin developed by Siemens Healthineers is one example of how medical device makers are using artificial intelligence (AI) to help doctors make more precise diagnoses as medicine enters an increasingly personalized age. Siemens Healthineers has built up a vast database of more than 250 million annotated images, reports and operational data on which to train its new algorithms. In the example of the digital twin, the AI system was trained to weave together data about the electrical and physical properties and the structure of a heart into a 3D image.

We present a novel framework that enables efficient probabilistic inference in large-scale scientific models by allowing the execution of existing domain-specific simulators as probabilistic programs, resulting in highly interpretable posterior inference. Our framework is general purpose and scalable, and is based on a cross-platform probabilistic execution protocol through which an inference engine can control simulators in a language-agnostic way. We demonstrate the technique in particle physics, on a scientifically accurate simulation of the tau lepton decay, which is a key ingredient in establishing the properties of the Higgs boson. High-energy physics has a rich set of simulators based on quantum field theory and the interaction of particles in matter. We show how to use probabilistic programming to perform Bayesian inference in these existing simulator codebases directly, in particular conditioning on observable outputs from a simulated particle detector to directly produce an interpretable posterior distribution over decay pathways. Inference efficiency is achieved via inference compilation where a deep recurrent neural network is trained to parameterize proposal distributions and control the stochastic simulator in a sequential importance sampling scheme, at a fraction of the computational cost of Markov chain Monte Carlo sampling.

We investigate how generative adversarial nets (GANs) can help semi-supervised learning on graphs. We first provide insights on working principles of adversarial learning over graphs and then present GraphSGAN, a novel approach to semi-supervised learning on graphs. In GraphSGAN, generator and classifier networks play a novel competitive game. At equilibrium, generator generates fake samples in low-density areas between subgraphs. In order to discriminate fake samples from the real, classifier implicitly takes the density property of subgraph into consideration. An efficient adversarial learning algorithm has been developed to improve traditional normalized graph Laplacian regularization with a theoretical guarantee. Experimental results on several different genres of datasets show that the proposed GraphSGAN significantly outperforms several state-of-the-art methods. GraphSGAN can be also trained using mini-batch, thus enjoys the scalability advantage.