Sony Broadcast & Professional Research Laboratories

About: Sony Broadcast & Professional Research Laboratories is a company organization based out in Taipei, Taiwan. It is known for research contribution in the topics: Signal & Image processing. The organization has 38708 authors who have published 63864 publications receiving 865637 citations.

Dislocation related issues in the degradation of GaN-based laser diodes

Abstract: We investigate degraded GaN-based laser diodes (LDs) on epitaxial lateral overgrown GaN layers in terms of dislocations. Almost all of the threading dislocations that appear in the wing regions are a-type dislocations. Their origins are the lateral extension of dislocations from the seed regions that contingently bend upwards to the episurface. Comparing short-lived LDs and long-lived LDs that have almost the same power consumption, we find that the relative levels of dislocation densities in their respective active layers are different. In the degraded LDs, neither dislocation multiplication from the threading dislocations nor any structural changes of the threading dislocations are observed. This indicates that degradation is not caused by dislocation multiplication at the active layers, which is usually observed in LDs featuring zincblende-based structures. The degradation rate is almost proportional to the square root of the aging time. Our results indicate that degradation is governed by a diffusion process, and a detailed degradation mechanism is proposed.

Reducing Wrong Labels in Distant Supervision for Relation Extraction

Abstract: In relation extraction, distant supervision seeks to extract relations between entities from text by using a knowledge base, such as Freebase, as a source of supervision. When a sentence and a knowledge base refer to the same entity pair, this approach heuristically labels the sentence with the corresponding relation in the knowledge base. However, this heuristic can fail with the result that some sentences are labeled wrongly. This noisy labeled data causes poor extraction performance. In this paper, we propose a method to reduce the number of wrong labels. We present a novel generative model that directly models the heuristic labeling process of distant supervision. The model predicts whether assigned labels are correct or wrong via its hidden variables. Our experimental results show that this model detected wrong labels with higher performance than baseline methods. In the experiment, we also found that our wrong label reduction boosted the performance of relation extraction.

Information processing apparatus, information processing method, and computer program

Abstract: The present invention is intended to provide an apparatus and a method for realizing the presentation of a content list and the reproduction of content in accordance with proper user preference on the basis of date and time at which content is viewed. Content preference values that change with time are computed. On the basis of the computed content preference values, a content list is generated for presentation to the user. For example, preference values are computed on the basis of daily time zone, holiday, weekday, and day-of-the-week and a content list in accordance with the computed preference values is generated for presentation to the user. The novel configuration allows the presentation of an optimum content list on the basis of date and time at which content is viewed, thereby realizing the selection and viewing of content that properly reflects user preference.

Wireless power supplying system

Abstract: Disclosed herein is a wireless power supplying system, including a power transmission device adapted to transmit power supplied thereto, a repeater device adapted to repeat the transmission power of the power transmission device, and a power reception device adapted to receive the power repeated by said repeater device.

3D virtual reality multi-user interaction with superimposed positional information display for each user

Abstract: Positions of users other than a particular user in a virtual reality space shared by many users can be recognized with ease and in a minimum display space The center (or intersection) of a cross of radar map corresponds to the particular user and the positions of other users (to be specific, the avatars of the other users) around the particular user are indicated by dots or squares colored red for example in the radar map This radar map is displayed on virtual reality space image in a superimposed manner