Hasil Pencarian  ::  Simpan CSV :: Kembali

"With recent advances in the basic analysis of speaker recognition, the technology can be expanded to a new system of audio event detection. The analysis of audio events is important in a variety of applications including audio surveillance, sports highlights and hearing disability support. Past project has given many challenges regarding to the detection and recognition speaker. This report presents an audio event detection method by using the Maximum Likelihood techniques. The algorithm uses Gauss/an Mixture Model (GMM) to provide a model of several types of sound. The Maximum Likelihood methods will give an estimation of all the parameters of the Gauss/an Mixture Model that can be used to identify what event(s) happen in audio signals. The focus of this work is on the ability of modelling different types of audio files and identifying what events occur regarding to the models. A complete experimental evaluation of the Gauss/an Mixture Model is conducted on a 150 speaker, 3 different types of sound, with each type of sound consisting of 10 audio files. The approach of Expectation Maximisation algorithm is applied in order to improve the performance of the classifier."

"This paper presents an audio event detection method by using the Maximum Likelihood techniques. The project's algorithm uses Gaussian Mixture Model (GMM) to provide a model of several types of sound. The Maximum Likelihood methods will give an estimation of all the parameters of the Gaussian Mixture Model that can be used to identify what event(s) happen in audio signals. The focus of this work is on the ability of modelling different types of audio files and identifying what events occur regarding the models that were previously constructed. A complete experimental evaluation of the Gaussian Mixture Model is conducted on 150 audio files under 12 particular conditions for the training process. And 15 audio files under the same 12 conditions for the identification process. Different audio files will be used for the detection process where each audio file consist several events. In order to improve the performance of the classifier, Expectation Maximization algorithm is applied. From the experimental evaluation that has been done, it can be concluded that the accuracy could be improved slightly through the trial number of the convergence criteria (LLH Convergence) and the number of mixture. The programming of this project is using MATLAB Version 7.0.1 and also using an additional Toolbox from Voicebox."

"ABSTRAKPendeteksian Kejadian Luar Biasa (KLB ) membutuhkan metode untukmendeteksi kejadian dalam waktu yang cepat agar KLB bisa ditanggulangi sedinimungkin. Salah satu cara yang bisa dilakukan untuk mempercepat pendeteksianKLB adalah dengan mengamati indikator-indikator dari KLB itu sendiri, sepertimengamati gejala-gejala dari suatu wabah penyakit. Indikator-indikator tersebutdiamati sebagai dataset. Dalam mendeteksi KLB juga ingin diketahui dimana danberapa lama KLB telah terjadi. Pada metode ini, tiga aspek diatas (Dataset,Lokasi, dan Waktu) diamati secara simultan melalui pendekatan Subset Scanyang mendeteksi KLB dengan melakukan pencarian terhadap kombinasi subsetsubsethimpunan dari tiga aspek tersebut. Oleh karena jumlah subset meningkatsecara eksponensial seiring bertambahnya jumlah anggota himpunan, dilakukanpereduksian subset dengan menggunakan sifat Linear Time Subset Scanning agarefisien secara komputasi. Sehingga, fast subset scan berarti mendeteksi KLBdengan waktu yang lebih cepat dan efisien secara komputasi. Sebagai ilsutrasi,dilakukan simulasi pendeteksian yang menggunakan data sintetis denganmengambil penyakit Chikungunya dan 2 kecamatan di Kota Depok sebagaiobjeknya.

---

ABSTRACTEvent Detection requires a method that can detect events in a short time sothat outbreaks can be addressed as early as possible. One way that can bedone to speed up the detection of outbreaks is to track indicators ofthe outbreak itself, such as observing the symptoms of a disease outbreak. Theindicators are observed as the datasets. In detecting outbreaks also wantto know where and how long outbreaks have occurred. In thismethod, three aspects above (Data Set, Location, and Time) isobserved simultaneouslywith Subset Scan approach that detects outbreaks by searching for thecombinations of subsets of the of three sets aspects. Because the numberof subsets increases exponentially by increasing number of members of theset, a reduction of subset is done using theLinear Time Subset Scanning properties that computationallyefficient. So fast subset scan means time to detect outbreaksis faster and computationally efficient. As ilsutration, performed detectionsimulations using data synthetic by taking Chikungunya disease and 2 districts inDepok as its object."

"Sistem kendali merupakan hal penting di dalam perancangan sebuah alat bantu berjalan untuk pasien pasca stroke yang mengalami hemiparetik pada kakinya. Sistem kendali yang baik harus mampu mengetahui keinginan bergerak atau berjalan dari manusia dan menerjemahkan keinginan tersebut menjadi sebuah gerakan yang alami melalui alat bantu berjalan yang umumnya digerakkan oleh sebuah perangkat DC motor. Sudah banyak penelitian yang telah dilakukan untuk melakukan deteksi terhadap keinginan manusia untuk bergerak atau berjalan melalui berbagai macam sensor yang dipasang pada otot-otot yang terkait. Fokus dalam penelitian ini adalah melakukan deteksi gaya berjalan melalui sinyal elektromiografi yang diperoleh dengan menggunakan sensor-sensor EMG yang dipasangkan pada permukaan 12 otot yang sangat berkaitan dengan gerakan atau gaya berjalan pada manusia. Adapun 12 otot ini terdiri dari 2 otot bahu yaitu Deltoid Anterior (DA) dan Deltoid Posterior (DP), dan 10 otot kaki yang terdiri dari Rectus Femoris (RF), Biceps Femoris (BF), Vastus Medialis (VM), Vastus Lateralis (VL), Tibialis Anterior (TA), Medial Gastrocnemius (MG), Soleus (S), Gluteus Maximus (GMax), Semitendinosus (ST), dan Peroneus Longus (PL). Sinyal elektromiografi dari 12 otot tersebut direkam dari 2 pasien sehat yang tidak mengalami gangguan berjalan, terdiri dari 1 orang pria dan 1 orang wanita. Sinyal tersebut kemudian diproses melalui aplikasi Matlab untuk dilakukan proses klasifikasi dengan menggunakan teknik Artificial Neural Network (ANN). Di samping itu, metode machine learning juga dilakukan yaitu dengan teknik Linear Discriminant Analysis (LDA), Support Vector Machine (SVM) dan K-Nearest Neighbor (KNN), yang bertujuan untuk mendapatkan perbandingan berbagai teknik tersebut agar didapatkan hasil dengan tingkat akurasi terbaik di dalam melakukan deteksi gaya berjalan yang dibedakan menjadi 3 yaitu: berjalan normal, naik tangga dan turun tangga. Hasil terbaik yang diperoleh dari penelitian ini dengan menggunakan algoritma ANN yang mampu menghasilkan prediksi sempurna dengan tingkat akurasi 100%, kemudian tingkat akurasi terbaik yang diperoleh dengan metode machine learning masing-masing untuk algoritma SVM adalah sebesar 99.2%, algoritma KNN sebesar 98.8% dan algoritma LDA sebesar 97.2% yang semuanaya diperoleh dari dataset kombinasi sinyal EMG otot bahu dan kaki. Hasil ini sangatlah penting di dalam penelitian yang akan dilakukan di kemudian hari dalam merancang sebuah sistem kendali yang mampu mengenali keinginan bergerak atau berjalan manusia baik saat berjalan normal maupun ketika hendak naik atau turun tangga sehingga alat bantu berjalan yang dihasilkan dapat digunakan dengan nyaman dan aman oleh pemakainya.

---

Control strategy is a fundamental role and very important part to create a walking assistive device for patients after stroke with a hemiparetic leg. A good control strategy must have the ability to predict the human motion or walking intention and naturally deliver force by the walking assistive device thereafter. This force is usually generated by the electric actuator using direct-drive motor. Recently, many studies have addressed and put more interest in predicting the human motion intention through various sensors which put on the surface of related skeletal muscles. This study focuses on gait event detection using electromyography signals from 12 muscles comprise of 2 shoulder muscles those are Deltoid Anterior (DA) and Deltoid Posterior (DP) and 10 lower limb muscles those are Rectus Femoris (RF), Biceps Femoris (BF), Vastus Medialis (VM), Vastus Lateralis (VL), Tibialis Anterior (TA), Medial Gastrocnemius (MG), Soleus (S), Gluteus Maximus (GMax), Semitendinosus (ST), and Peroneus Longus (PL). The EMG signals are recorded unilaterally using surface EMG sensor from 2 healthy subjects without walking disorder, consist of 1 male and 1 female. The signals are processed on Matlab platform subsequently for classification process using Artificial Neural Network (ANN) technique. Besides, the machine learning methods are also used in this research i.e. Linear Discriminant Analysis (LDA), Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). The purpose of using several methods is to output the comparison with highest accuracy result in predicting the gait events which are divided into 3 types: normal walking, stair ascent, and stair descent. The best outcome along this research is generated from ANN algorithm which could steadily predict without any error with accuracy rate 100%. Furthermore, the best results from machine learning method are 99.2% using SVM algorithm, 98.8% using KNN algorithm and 97.2% using LDA algorithm. All those performances are resulted from datasets with combination between EMG signals from shoulder and lower limb muscles. This achievement becomes a significant factor for the future studies to design a control strategy with good human-robot interaction that can recognize the human motion intention in each different gait event to contrive comfort and safety walking assistive device for the wearer. "