Machine Learning Methods for Trustworthy Autonomous Systems (video)

There is a fast development of different machine learning methods – for object classification, tracking, action recognition and other tasks with multiple types of data – from images and videos to data from wireless sensor networks. Autonomous image and video analytics face a number of challenges due to the huge volumes of data that sensors provide, the changeable environmental conditions and other factors. However, it is important to know when the methods work well and when they are not reliable, e.g. how much could we trust the obtained results? How could we characterize trust is a related question. How could we quantify the impact of uncertainties on the developed solutions? This talk will discuss current trends in the area of machine learning and show results for image and video analytics for autonomous systems. Recent advance of data-driven and model-based approached have led to increasing the level of autonomy in a number of areas. Gaussian Process methods are one type of Machine Learning methods that have proven their power to solve for such tasks for autonomous systems. This talk will present recent results on tracking, automated video analytics especially with Gaussian process methods and other methods, with their pros and cons. Some of these results are part of Digital twins, recently developed new tools that incorporate machine learning and artificial intelligence methods. The presenter will also discuss the big potential of Digital Twins, the opportunities and challenges that they bring. Other related applications will be discussed such and autonomous fire detection and robot pipe networks inspection.