More than an actual talk, this will be an informal exchange on the theme of "A potential paradigm shift in vision search." Active participation is warmly welcomed. This is an internal event, meant for SCIoI members only.
With research in machine learning (ML) and natural language processing (NLP), we aim to give machines the ability to understand and use human language. In this talk, I give a high level introduction of some of the challenges of the field and give an overview of basic NLP tasks (and show some demos). I also
Patients suffering from schizophrenia are less susceptible to various perceptual illusions (and to some hallucinations, too) than most healthy individuals. Yet, schizophrenia patients’ perception-forming processes have been characterised as aberrant, as producing false inferences and irrational mental states. This characterisation is consistent with the idea that perceptual experiences and processes can be appraised as rational
Reward-based Learning and Decision Making under Risk Reinforcement learning provides a framework for making agents learn policies through feedback signals (“rewards”), which provide information about whether their actions or action sequences were successful or not. Reinforcement learning also provides a framework for understanding how humans learn and decide given reward information only. Standard reinforcement learning
Abstract: Research in robotics and A.I. aims at optimizing very specific task rewards. Intelligent animals have a high degree of curiosity, and recent results have shown that instrumental reward optimization is a poor explanation for their behavior. We can show that to explain empirical results from animals, we need to have the drive to optimize
Abstract: I would like to give a personal perspective of the scientific motivation and framing of SCIoI and relate them to the research of my lab, the Robotics and Biology Laboratory. But at the same time, I would like to critically question and discuss all of these things, in an attempt to move towards a
Rainer Mühlhoff and John-Dylan Haynes participated in a fishbowl discussion on predictive AI.
Abstract: We typically have a clear idea about the final design and functionality of a robot before we start building it. We apply this top-down design approach to a wide range of robotic systems and it allows our robots to be more optimized, autonomous, and programmable. However, if we want to design and actuate multiple
Curiosity and information demand: how we can study them and why we should care A rapidly growing literature has recently emphasized the importance of our sense-making instincts, including complex investigative behaviors such as curiosity, for behavior and brain function. While much of this literature has focused on simple forms of decision making, we explored its
Ecologically rational decision making How do we make inferences about a world full of uncertainty and given the mind’s natural bounds in computational abilities? I present a perspective according to which the decision maker is equipped with a repertoire of strategies, containing both simple heuristics and more complex strategies that are adaptive under different ecological
Abstract: Deep neural networks underlie many state of the art solutions to hard problems in computer vision, natural language processing or playing Go. Yet, their power comes with a price. Deep networks transform inputs gradually into outputs, using many parameters and intermediary activations. Understanding what a network has learned, how inputs are mapped to outputs,
Abstract Most cognitive sciences (including AI) appeal to representations in explaining (or trying to create) systems capable of cognition, and especially of complex, intelligent behaviour. In philosophy, considerable attention has been dedicated to the problem of explaining in scientifically-acceptable ways how representations come to represent what they do. In contrast, there has been much less
Marchstr. 23 – MAR 5-2
10587 Berlin
Email: info@scioi.de
Tel: +49 30 314 70660
