Deep learning has become a mainstream method of hyperspectral image (HSI) classification. Many DL-based methods exploit spatial-spectral features to achieve better classification results. However, due the complex backgrounds in HSIs, existing usually show unsatisfactory performance for class pixels located on land-cover category boundary area. In large part, this is because network susceptible interference by irrelevant information around target pixel training stage, resulting inaccurate...

The probability that a user will click search result depends both on its relevance and position the results page. based model explains this behavior by ascribing to every item an attraction probability, examination probability. To be clicked, must attractive examined. probabilities of item-position pair being clicked thus form entries rank-1 matrix. We propose learning problem Bernoulli bandit where at each step, agent chooses row column arms, receives product their Bernoulli-distributed...

A search engine recommends to the user a list of web pages. The examines this list, from first page last, and clicks on all attractive pages until is satisfied. This behavior can be described by dependent click model (DCM). We propose DCM bandits, an online learning variant where goal maximize probability recommending satisfactory items, such as main challenge our problem that we do not observe which item satisfactory. computationally-efficient algorithm for solving problem, dcmKL-UCB;...

We study the online influence maximization problem in social networks under independent cascade model. Specifically, we aim to learn set of "best influencers" a network while repeatedly interacting with it. address challenges (i) combinatorial action space, since number feasible influencer sets grows exponentially maximum influencers, and (ii) limited feedback, only influenced portion is observed. Under stochastic semi-bandit propose analyze IMLinUCB, computationally efficient UCB-based...

Objectives Optic neuritis is (ON) believed to be an immune-mediated disease; however, the association between optic and autoimmune diseases remains unclear. This study aimed identify incidence rate adjusted hazard ratio (aHR) of in patients with neuritis. Methods nationwide, population-based, retrospective cohort collected patients’ data 1999 2013 from National Health Insurance Research Database Taiwan. A total 9,235 were included. Using 1:4 propensity scoring, 1,847 enrolled group 7,388...

We contribute to the ongoing research on external (disturbance) response of feedback-controlled nonlinear systems by further exploring disturbance a canonical system, namely double-integrator subject actuator saturation. Precisely, we aim delineate classes bounded, persistent disturbances that do or not drive system's state unbounded. Our key result is signal whose integral bounded over all intervals - which can be offset such its cannot The class signals includes periodic and almost-periodic ones.

This paper presents the problem of robustness controller design for requirements technological choice regulation. The uncertainty regulation model is studied. theory control applied to analysis above-mentioned model. Simulation results show that potential energy can increase in accordance with prediction object system and external restrained effectively process telecommunication

The purpose of this paper is to track the Regional Economic Development dynamically, and realize its fuzzy control. This presents a new integration method. method is, firstly, Fuzzy Evaluation control rule based on lots statistical data expert's experience was proposed; secondly, Controller designed. Finally, simulation carried out by applying MATLAB GUI Simulink. result shows that can trace Competitiveness, Control only 5 rules, acquire better than Classical Method evidently.

The main research methods of this paper are entropy weight method and simulation analysis method. This can be totally divided into five steps: the first step is to finish building evaluation index system telecommunication industries' core competence, second modeling work, third debugging running program, forth get outcome last finding out related solution measures schemes.

Programmable Automation Controller (PAC) is applied widely in various fields of engineering due to its high reliability and process speed as well convenience. Aimed at the characteristics PAC software project based on PLC control, some program design methods key technique are introduced realize intelligent which makes use ladder language. The frame structure program, sampling data, checking list calculation deviation change rate expatiated this paper.

The paper aims to study the effect of core-competitiveness-mimicking behaviors and their features China's three major telecommunications companies: China Mobile Communication Group Corporation (CMCC), United Telecommunications (CU) Network Communications (CNC) under certain rules dynamically continuously.