著者
片岡 駿
出版者
一般社団法人 電子情報通信学会
雑誌
電子情報通信学会 基礎・境界ソサイエティ Fundamentals Review (ISSN:18820875)
巻号頁・発行日
vol.11, no.4, pp.256-265, 2018-04-01 (Released:2018-04-01)
参考文献数
20

本稿では,マルコフ確率場を用いた生成モデルの方法について解説を行う.マルコフ確率場とは確率変数間の関係性をグラフ構造で表すことができる確率分布の総称であり,生成モデルの方法は確率分布を用いてデータセットの発生源を模倣する統計的機械学習法の一つである.まず,確率変数間の関係性をグラフ構造で表現するマルコフ確率場とその性質について概説し,確率分布に対する統計的機械学習法の一つである生成モデルの方法について述べる.その後,マルコフ確率場に対する基本的な学習法について説明し,実際の学習で役立つ幾つかのアルゴリズムを紹介する.また,マルコフ確率場を用いた生成モデルの方法の例として,制限ボルツマン機械について紹介する.
著者
林 初男
出版者
一般社団法人 電子情報通信学会
雑誌
電子情報通信学会 基礎・境界ソサイエティ Fundamentals Review (ISSN:18820875)
巻号頁・発行日
vol.8, no.2, pp.102-114, 2014-10-01 (Released:2014-10-01)
参考文献数
50

工学でも大きな興味が持たれている脳のニューロンや神経回路網は非線形なシステムである.ニューロンは規則的に発火するだけでなく,膜電位あるいは刺激の強さや周波数に依存して不規則に発火することがある.この不規則な発火はカオスと呼ばれる決定論的な力学則に従う非周期振動である.神経回路網の場合は,ニューロンの集団活動を反映した電場電位が周期刺激に対して引き込みやカオス応答を起こす.すなわち,ニューロン集団の同期の程度や同期したニューロン数が刺激に応じて変化する.本稿では,非線形ダイナミクスの観点から脳を概観し,脳の情報処理機構に関する研究の一例として,海馬における確率共鳴と記憶パターンの想起について述べる.
著者
Wei Jizeng Chang Yisong Li Bingchao Guo Wei Sun Jizhou
出版者
一般社団法人 電子情報通信学会
雑誌
IEICE Electronics Express
巻号頁・発行日
vol.12, no.10, pp.20150314, 2015

The traditional post-TnL vertex cache (abbr. 'post-VC') in embedded GPUs (EGPUs) with only one vertex or unified shader does not fit to multi-shader EGPUs for two reasons. As multiple shaders run in parallelism, (a) the out-of-order vertex processing may raise the post-VC inconsistency that leads to cache the error data, and (b) it is very hard to detect in time which vertices are saved in the post-VC in the stage of vertex fetching, resulting in the low performance. In this paper, we propose a modified post-VC including a decoupling cache and a vertex batch in-order commit controller, which can guarantee that the data SRAM and index tag can be updated in-order according to the same replacement policy in the different stages of vertex processing. The function of the proposed post-VC is verified on a FPGA-based platform. Experimental results show that it increases the performance by an average of 172% and 80.6% compared to the EGPU without/with the traditional post-VC respectively at a little expense.
著者
VIDMAR Luka PESKO Marko ŠTULAR Mitja PETERNEL Blaž KOS Andrej POGAČNIK Matevž
出版者
一般社団法人 電子情報通信学会
雑誌
IEICE Trans. Commun. (ISSN:09168516)
巻号頁・発行日
vol.98, no.9, pp.1906-1913, 2015

User context and user location in particular play an important role in location-based services (LBS). The location can be determined by various positioning methods. These are typically evaluated with average positioning error or percentile values, which are not the most suitable metrics for evaluation of how a positioning method functions in the semantic space. Therefore, we propose a new method for evaluation of positioning accuracy in the semantic space. We focus on two types of semantic user locations that are widely available in urban areas: the street address and the categories of the surrounding points of interest (POIs). We demonstrate its use on ten different positioning methods: a standalone satellite navigation device, GPS module on a smartphone, two versions of Foursquare positioning service, Google positioning service, a positioning service of the local mobile operator, and four other possible variants of mobile operator-based positioning methods. The evaluation suggests that approach with the street addresses is more promising approach due to either sparse or unevenly distributed POIs. Furthermore, some of the positioning methods that are less accurate in Euclidean space, such as a combination of the GPS data with the mobile operator-based method that relies on the propagation models, performed comparably well in the semantic space as the methods that are using more accurate technologies, such as Google and Foursquare.
著者
MATSUBARA Shigeo WANG Meile
出版者
一般社団法人 電子情報通信学会
雑誌
IEICE Transactions on Information and Systems
巻号頁・発行日
vol.97, no.9, pp.2415-2422, 2014
被引用文献数
1

We propose a method for finding an appropriate setting of a pay-per-performance payment system to prevent participation of insincere workers in crowdsourcing. Crowdsourcing enables fast and low-cost accomplishment of tasks; however, insincere workers prevent the task requester from obtaining high-quality results. Instead of a fixed payment system, the pay-per-performance payment system is promising for excluding insincere workers. However, it is difficult to learn what settings are better, and a naive payment setting may cause unsatisfactory outcomes. To overcome these drawbacks, we propose a method for calculating the expected payments for sincere and insincere workers, and then clarifying the conditions in the payment setting in which sincere workers are willing to choose a task, while insincere workers are not willing to choose the task. We evaluated the proposed method by conducting several experiments on tweet labeling tasks in Amazon Mechanical Turk. The results suggest that the pay-per-performance system is useful for preventing participation of insincere workers.