Every government has a requirement to ensure that laws are not only equitable to all citizens but also applicable. Philosophers for centuries have argued and debated about the relationship of the individual in a society, and the concept of fairness and equality is generally a main driving force in democratic populations. As we've seen with government polarization, laws can become really slow to get adopted. Usually, elected officials pass a law to assign a set of responsibilities to an agency or to a department. This responsible body can update the regulations as they see fit for the duration of its mandate.

Google's AI researchers recently showed off a new method for teaching computers to understand why some images are more aesthetically pleasing than others. Traditionally, machines sort images using basic categorization – like determining whether an image does or does not contain a cat. The new research demonstrates that AI can now rate image quality, regardless of category. The process, called neural image assessment (NIMA), uses deep learning to train a convolutional neural network (CNN) to predict ratings for images. Our approach differs from others in that we predict the distribution of human opinion scores using a convolutional neural network … Our resulting network can be used to not only score images reliably and with high correlation to human perception, but also to assist with adaptation and optimization of photo editing/enhancement algorithms in a photographic pipeline.

In this paper, we focus on learning intelligent agents through model-free reinforcement learning. Rather than arguing that reinforcement learning is the right abstraction for attaining intelligent behavior, we consider the issue of finding useful abstractions to represent the agent and the environment when verification is in order. Indeed, verifying that the agent’s behavior complies to some stated safety property — an ”Asimovian” perspective — only adds to the challenge that abstracting intelligence represents per se. In the paper, we show an example application about verification of abstractions in model-free learning, and we argue about potential (more) useful abstractions in the same context.

Although planning for the tasks a household robot has to perform appears to be easy, there exists the problem that the robot is usually uncertain about the state of the household when starting to plan. For example, when getting the order of tidying up the kitchen, the robot does not know what objects it will have to put away and whether there are actually any objects that need to be put away. Furthermore, while sensing operations can provide moreinformation about the environment, things can go wrong when executingan action. In this paper, we try to identify conditions under which classical planning can be used in a replanning loop in order to solve the planning problem in nondeterministic partially observable open domains. In particular, we will define completeness and soundness of replanning with respect to nondeterministic planning and we will identify a PSPACE-checkable condition that guarantees soundness.

We provide a worst-case analysis of selective sampling algorithms for learning linear threshold functions. The algorithms considered in this paper are Perceptron-like algorithms, i.e., algorithms which can be efficiently run in any reproducing kernel Hilbert space. Our algorithms exploit a simple margin-based randomized rule to decide whether to query the current label. We obtain selective sampling algorithms achieving on average the same bounds as those proven for their deterministic counterparts, but using much fewer labels. We complement our theoretical findings with an empirical comparison on two text categorization tasks. The outcome of these experiments is largely predicted by our theoretical results: Our selective sampling algorithms tend to perform as good as the algorithms receiving the true label after each classification, while observing in practice substantially fewer labels.

We provide a worst-case analysis of selective sampling algorithms for learning linear threshold functions. The algorithms considered in this paper are Perceptron-like algorithms, i.e., algorithms which can be efficiently run in any reproducing kernel Hilbert space. Our algorithms exploit a simple margin-based randomized rule to decide whether to query the current label. We obtain selective sampling algorithms achieving on average the same bounds as those proven for their deterministic counterparts, but using much fewer labels. We complement our theoretical findings with an empirical comparison on two text categorization tasks. The outcome of these experiments is largely predicted by our theoretical results: Our selective sampling algorithms tend to perform as good as the algorithms receiving the true label after each classification, while observing in practice substantially fewer labels.

I am absolutely delighted to be able to join in this morning to offer my reflections on the occasion of the official beginning of the Robotics Institute. Beginnings are full of promise and potential. This one is no exception. What the Robotics Institute will become -- what effects it will have, both witting and unwitting -- are for the future to tell. What we all have now is a sense of adventure and anticipation.