Search

With increasingly complex neural network architectures and heterogeneous device characteristics, finding a reasonable graph partitioning and device placement strategy is challenging. There have been prior attempts at learned approaches for solving device placement, these approaches are computationally expensive, unable to handle large graphs consisting over 50000 nodes, and do not generalize well to unseen graphs. To address all these limitations, we propose an efficient single-shot, generalized deep RL method (SGDP) based on a scalable sequential attention mechanism over a graph neural network that is transferable to new graphs. On a diverse set of representative deep learning models, our method on average achieves 20% improvement over human placement and 18% improvement over the prior art with 15× faster convergence. We are the first to demonstrate super human performance on 8-layer recurrent neural network language model and 8-layer GNMT consisting of over 50000 nodes, on 8-GPUs. We provide rationales and sensitivity study on model architecture selections. [ABSTRACT FROM AUTHOR]

We proposed and experimentally demonstrated a cost-efficient and easy-to-fabricate flexible thin-film nanocrystal quantum dot (NCQD) photodetector, by sandwiching a piece of NCQD-soaked tracing paper between two PEDOT:PSS electrodes. The highly conductive and transparent multilayer polymer electrodes are printed with an office inkjet printer without further patterning on commercial unmodified transparency films. The tracing paper is able to act as both the bed for NCQD attachment and the interspacing layer to prevent short-circuiting in vertical devices. Consistent increase in device performance is observed when the bending radius becomes smaller. Linear I-V characteristics are obtained, which show our device structure works as a photoconductor. [ABSTRACT FROM PUBLISHER]

Abstract: Since its invention, paper has become one of the main carriers of our cultural, scientific, political, economic and historical information. Given the importance of this material, its preservation is a matter of great interest. Paper can be deteriorated due to physical, chemical and biological agents. Within microorganisms, fungi are the major paper biodeteriogens. Throughout history, several methods have been used to prevent and stop fungal deterioration on paper based materials. In this work we present a review of the main chemical and physical methods used to avoid fungal paper biodeterioration until nowadays and also of some new approaches tested recently. The advantages and disadvantages of these methods are discussed as well as their health effects. Studies regarding antifungal compositions, methods of application, performance and effects on the treated materials are also presented with the aim of providing a clear set of conclusions on the topic. [Copyright &y& Elsevier]

Algorithm performance evaluation is so entrenched in the machine learning community that one could call it an addiction. Like most addictions, it is harmful and very difficult to give up. It is harmful because it has serious limitations. Yet, we have great faith in practicing it in a ritualistic manner: we follow a fixed set of rules telling us the measure, the data sets and the statistical test to use. When we read a paper, even as reviewers, we are not sufficiently critical of results that follow these rules. Here, we will debate what are the limitations and how to best address them. This article may not cure the addiction but hopefully it will be a good first step along that road. [ABSTRACT FROM AUTHOR]

This article reports the results from an experiment examining the effect of paper color on student performance on quiz scores in Principles of Accounting courses. The ten-year study included classes taught by five instructors and involved almost 4,000 student scores. The paper colors utilized on the quizzes were four pastels (yellow, green, blue, and pink) and white. Contrary to some similar studies, the current study found that paper color did have a significant effect on student performance. Students with white and yellow colored paper performed significantly better than students with either blue or pink colored paper . Students with green colored paper performed somewhere in the middle of the other four colors. Prior research is compared with the current study findings, and it is suggested that, given the past somewhat inconsistent findings of color effects, further research is warranted to disentangle the mixed results and instructors should consider utilizing other means to deter cheating. If an instructor feels that the only method he or she can employ to monitor cheating is to color-code exams, the instructor should consider using a combination of white, yellow, or green pastel colored exams, and avoid blue or pink colored exams. [ABSTRACT FROM AUTHOR]