Set-membership filtering for automatic guided vehicles with unknown-but-bounded noises

This paper is concerned with the state estimation problem for an automatic guided vehicle (AGV). A novel set-membership filtering (SMF) scheme is presented to solve the state estimation problem in the trajectory tracking process of the AGV under the unknown-but-bounded (UBB) process and measurement noises. Different from some existing traditional filtering methods, such as Kalman filtering method and [Formula: see text] filtering method, the proposed SMF scheme is developed to provide state estimation sets rather than state estimation points for the system states to effectively deal with UBB noises and reduce the requirement of the sensor precision. Then, in order to obtain the state estimation ellipsoids containing the true states, a set-membership estimation algorithm is designed based on the AGV physical model and S-procedure technique. Finally, comparison examples are presented to illustrate the effectiveness of the proposed SMF scheme for an AGV state estimation problem in the present of the UBB noises.

Analysis of comparative performance of deep learning models for sentiment analysis

Sentiment analysis of text can be performed using machine learning and natural language processing methods. However, there is no single tool or method that is effective in all cases. The objective of this research project is to determine the effectiveness of neural network-based architecture to perform sentiment analysis of customer comments and reviews, such as the ones on Amazon site. A typical sentiment analysis process involves text preparation (of acquired content), sentiment detection, sentiment classification and analysis of results. In this research, the objective is to a) identify the best approach for text preparation in a given application (text filtering approach to remove errors in data), and, most importantly, b) what is the best machine learning (feed forward neural nets, convolutional neural nets, Long Short-Term Memory networks) approach that provides best classification accuracy. In this research, a set of three thousand two hundred reviews of food related products were used to train and experiment with a neural network-based sentiment analysis system. The neural network implementation of six different models provided close to one-hundred percent accuracy of test data, and a decent test accuracy in mid-80%. The results of the research would be useful to businesses in evaluating customer preferences for products or services.

Penerapan Metode Single Linkage dengan Manhattan Distance Similarity dalam Mengelompokkan Trens Topik Kerja Praktik

Data laporan judul kerja praktik (KP) biasanya hanya terkumpul di perpustakaan dan jarang dipubilkasikan ke mahasiswa, hal ini menyebabkan kesulitan bagi mahasiswa yang akan mengkasesnya. Berdasarkan permasalahan tersebut, maka dibuatlah suatu program pada penlitian ini untuk pengelompokkan Trend Topik. Metode yang digunakan dalam penelitian ini adalah Manhattan Distance Similariy dan Single Linkage. Sebelum masuk tahapan text mining, perlu dilakukan perancangan diantaranya perancangan basis data dan antar muka (interface). Tahapan dan text mining adalah mengumpulkan data (collect data), penguraian teks (text mining), penyaringan teks (text filtering), pembobotan kata (calculate term count), similarity, pengelompokan, dan pengujian. Hasil dari penelitian ini adalah program yang dapat mengolah data judul KP menjadi pola kelompok Trend Topik KP. Dari 905 data yang di dapatkan, terbentuk 7 kelompok yaitu Sistem Informasi, Multimedia, Jaringan, Web, Kewirausahaan, Magang, dan Pelatihan. Tetapi dari hasil pengujian Purity Test didapatkan nilai sebesar 0,267, yang artinya Manhattan Distance Similarity dan Single Linkage kurang cocok untuk mengelompokkan Judul KP.

User-Query Processing through Dynamic Tweets Status Recommender System

With increasing user information volume in online social networks, recommender systems have been an effective method to limit such information overload. The requirements of recommender systems specified, with widespread adoption in many internet social Twitter, Facebook, and Google online applications. In recent years, the micro-blogging in Twitter has brought greater importance to online users as a channel spreading knowledge and information. Through Twitter, users can find the relevant information on the search they perform, but understanding the past, present, and future information relevant to the investigation source is needed real-time information. Estimating the successful tweet status (history, ongoing, and prospective) among the huge population of Twitter members is important to satisfy the needs of Twitter online content readers. In this paper, a Dynamic Tweets Status Recommender System (DTSRS) is designed by creating a set of dynamic recommendations to a Twitter user based on usability, consisting of people who post tweets, which is exciting present and future. The proposed recommender system is implemented through two approaches: the first is to analyze the Twitter member online tweets, select and understand the content of that tweet, and the second predicts the understanding of the tweet content, suggest the dynamic status of the tweets. In this paper, the Twitter user tweets' views are expressed after examining the depth of content, different types of user interfaces, text filtering, and machine learning technique. The set of results through tweets experimentations with database operators carried out to evaluate and comparability the proposed recommender system's performance.

Event-triggered distributed filtering for Markov jump systems over sensor networks

This article deals with the distributed filtering problem for a class of discrete-time Markov jump systems over sensor networks. First, in the distributed filtering network, each local filter simultaneously fuses the estimation and measurement from itself and neighboring nodes to achieve the system state estimation. And each sensor intelligent node is embedded with an event-triggered sampling mechanism, which can reduce the computation load or saving limited network bandwidth. Then, we use Bernoulli stochastic variables to describe whether the filtering network can successfully receive the system jump modes. Next, based on the Lyapunov stability theory, we design a distributed filter dependent on partial system modes, which has the exponential stability in mean square and [Formula: see text] performance. Finally, all desired estimator parameters can be obtained by solving a set of linear matrix inequalities. Moreover, two numerical examples are given to illustrate the effectiveness of the distributed [Formula: see text] filtering design approach.