Carnegie Mellon University

About: Carnegie Mellon University is a education organization based out in Pittsburgh, Pennsylvania, United States. It is known for research contribution in the topics: Computer science & Robot. The organization has 36317 authors who have published 104359 publications receiving 5975734 citations. The organization is also known as: CMU & Carnegie Mellon.

A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning

Abstract: Sequential prediction problems such as imitation learning, where future observations depend on previous predictions (actions), violate the common i.i.d. assumptions made in statistical learning. This leads to poor performance in theory and often in practice. Some recent approaches (Daume III et al., 2009; Ross and Bagnell, 2010) provide stronger guarantees in this setting, but remain somewhat unsatisfactory as they train either non-stationary or stochastic policies and require a large number of iterations. In this paper, we propose a new iterative algorithm, which trains a stationary deterministic policy, that can be seen as a no regret algorithm in an online learning setting. We show that any such no regret algorithm, combined with additional reduction assumptions, must find a policy with good performance under the distribution of observations it induces in such sequential settings. We demonstrate that this new approach outperforms previous approaches on two challenging imitation learning problems and a benchmark sequence labeling problem.

Opportunities and obstacles for deep learning in biology and medicine.

Abstract: Deep learning describes a class of machine learning algorithms that are capable of combining raw inputs into layers of intermediate features. These algorithms have recently shown impressive results across a variety of domains. Biology and medicine are data-rich disciplines, but the data are complex and often ill-understood. Hence, deep learning techniques may be particularly well suited to solve problems of these fields. We examine applications of deep learning to a variety of biomedical problems-patient classification, fundamental biological processes and treatment of patients-and discuss whether deep learning will be able to transform these tasks or if the biomedical sphere poses unique challenges. Following from an extensive literature review, we find that deep learning has yet to revolutionize biomedicine or definitively resolve any of the most pressing challenges in the field, but promising advances have been made on the prior state of the art. Even though improvements over previous baselines have been modest in general, the recent progress indicates that deep learning methods will provide valuable means for speeding up or aiding human investigation. Though progress has been made linking a specific neural network's prediction to input features, understanding how users should interpret these models to make testable hypotheses about the system under study remains an open challenge. Furthermore, the limited amount of labelled data for training presents problems in some domains, as do legal and privacy constraints on work with sensitive health records. Nonetheless, we foresee deep learning enabling changes at both bench and bedside with the potential to transform several areas of biology and medicine.

Time Paths in the Diffusion of Product Innovations

Abstract: This study attempts to measure and analyse the diffusion of product innovations. Diffusion is defined as the spread in the number of producers engaged in manufacturing a new product. Thus, the term refers to the net entry rate in the market for a new product. We trace the history of diffusion for 46 new products and examine the inter-relations among diffusion, other aspects of technological change, price, output, and certain attributes of the relevant markets. To explain the 46 product histories, we construct a theory of the development of industries for new products. Our theory combines elements of traditional, neoclassical models with what Nelson and Winter (I974) have termed an evolutionary theory. A novel feature is that the historical sequence, or time path, of events is viewed as a critical determinant of the ultimate structure of new product markets. Thus the time path of events determines not only the course traversed in reaching the end result but the ultimate market structure itself. The paper is organised in four sections. In Section I we present our theory. In Section II we construct a series of alternative theories of the development of industries for new products based on approaches to be found in received literature. The evidence from the 46 new product histories is examined in Section III. Finally, a brief summary of principal findings follows in Section IV.

Cortical activation and synchronization during sentence comprehension in high-functioning autism: evidence of underconnectivity

Abstract: Summary The brain activation of a group of high-functioning autistic participants was measured using functional MRI during sentence comprehension and the results compared with those of a Verbal IQ-matched control group. The groups differed in the distribution of activation in two of the key language areas. The autism group produced reliably more activation than the control group in Wernicke’s (left laterosuperior temporal) area and reliably less activation than the control group in Broca’s (left inferior frontal gyrus) area. Furthermore, the functional connectivity, i.e. the degree of synchronization or correlation of the time series of the activation, between the various participating cortical areas was consistently lower for the autistic than the control participants. These findings suggest that the neural basis of disordered language in autism entails a lower degree of information integration and synchronization across the large-scale cortical network for language processing. The article presents a theoretical account of the findings, related to neurobiological foundations of underconnectivity in autism.