Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Proceeding of the Electrical Engineering Computer Science and Informatics

Narrow Window Feature Extraction for EEG-Motor Imagery Classification using k-NN and Voting Scheme Adi Wijaya; Teguh Bharata Adji; Noor Akhmad Setiawan
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 5: EECSI 2018
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (368.009 KB) | DOI: 10.11591/eecsi.v5.1665

Abstract

Achieving consistent accuracy still big challenge in EEG based Motor Imagery classification since the nature of EEG signal is non-stationary, intra-subject and inter-subject dependent. To address this problems, we propose the feature extraction scheme employing statistical measurements in narrow window with channel instantiation approach. In this study, k-Nearest Neighbor is used and a voting scheme as final decision where the most detection in certain class will be a winner. In this channel instantiation scheme, where EEG channel become instance or record, seventeen EEG channels with motor related activity is used to reduce from 118 channels. We investigate five narrow windows combination in the proposed methods, i.e.: one, two, three, four and five windows. BCI competition III Dataset IVa is used to evaluate our proposed methods. Experimental results show that one window with all channel and a combination of five windows with reduced channel outperform all prior research with highest accuracy and lowest standard deviation. This results indicate that our proposed methods achieve consistent accuracy and promising for reliable BCI systems.
Impact of Matrix Factorization and Regularization Hyperparameter on a Recommender System for Movies Gess Fathan; Teguh Bharata Adji; Ridi Ferdiana
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 5: EECSI 2018
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (336.245 KB) | DOI: 10.11591/eecsi.v5.1685

Abstract

Recommendation system is developed to match consumers with product to meet their variety of special needs and tastes in order to enhance user satisfaction and loyalty. The popularity of personalized recommendation system has been increased in recent years and applied in several areas include movies, songs, books, news, friend recommendations on social media, travel products, and other products in general. Collaborative Filtering methods are widely used in recommendation systems. The collaborative filtering method is divided into neighborhood-based and model-based. In this study, we are implementing matrix factorization which is part of model-based that learns latent factor for each user and item and uses them to make rating predictions. The method will be trained using stochastic gradient descent with additional tricks and optimization of regularization hyperparameter. In the end, neighborhood-based collaborative filtering and matrix factorization with different values of regularization hyperparameter will be compared. Our result shows that matrix factorization method with lowest regularization hyperparameter outperformed the other methods in term of RMSE score. In this study, the used functions are available from Graphlab and using Movielens 100k data set for building the recommendation systems.