Despite reaching state-of-the-art performance in relation to key machine learning tasks, deep neural networks are still nowhere close in bridging the gap with respect to the range of essential cognitive abilities associated with human-level intelligence. Although scale has proven to be a definitive driver in building increasingly better performing models,  neural nets still exhibit serious vulnerabilities and erroneous behavior in terms of brittleness, spurious correlations, lack of interpretability, or more recently, hallucinating misinformation.  In this talk I will set-out by identifying a few key underlying. limitations in the deep learning literature, as well as addressing shortcomings of machine learning techniques more broadly. Following, I will point towards several research directions by reflecting on contributions from my previous and ongoing research. The presentation will focus on approaches towards mitigating some of these identified shortcomings, showcased across various application domains including computer vision, natural language processing and internet of things settings.

Dr. Radu-Casian Mihailescu is an Associate Professor in Computer Science at the Faculty Mathematical and Computer Sciences, in Heriot-Watt University (UK). His research focuses primarily on advances in the field of machine learning (ML), with particular emphasis on state-of-the-art deep learning architectures. Key areas of his research include topics such as out-of-distribution generalization, transfer learning, meta-learning, active learning, interactive learning, few-shot learning, domain adaptation, machine understanding, as well as building upon distributed representations learned by deep networks and incorporating reasoning as an integral part of the learning procedure. Moreover, he is also active in numerous applications of ML approaches to various real-world case-studies such as computer vision, natural language processing for fake news detection, activity recognition based on the internet of things infrastructures, context-adaptive surveillance systems or ambulance coordination for acute stroke care. He is also affiliated with the Internet of Things and People Research Center and has previously acted as Program Director for Applied Data Science Master’s Degree at Malmö University in Sweden. He has participated in over 10 national and international European research projects, published over 50 academic papers and has been serving as a program committee member and reviewing in a number of international scientific conferences, journals and workshops.