Hasil Pencarian  ::  Simpan CSV :: Kembali

Hasil Pencarian

Ditemukan 98918 dokumen yang sesuai dengan query
cover
Carlo Johan Nikanor
"Perkembangan pesat teknologi telah memberikan akses kepada masyarakat untuk mengemukakan opini dan evaluasi pribadi di media sosial dan berbagai penjuru dunia digital. Hal ini menjadi pemicu berkembangnya ilmu analisis sentimen atau sering disebut juga opinion mining yang merupakan pengaplikasian dari ilmu machine learning. Umumnya, metode machine learning mempelajari satu domain untuk menghasilkan suatu model, tetapi dengan pengembangan lanjut dihasilkan lifelong learning dimana pembelajaran model berlangsung secara kontinu menggunakan berbagai source domain. Pada tahun 2022, Osmardifa melakukan penelitan mengenai perbandingan kinerja model Bidirectional Encoding Representation from Transformers (BERT) terhadap kinerja model Convolutional Neural Network (CNN) dan model Long Short-Term Memory (LSTM) untuk lifelong learning. Namun, dari perbandingan kinerja tersebut hanya menggunakan satu kombinasi urutan domain dari total 120 kombinasi dari urutan 5 source domain. Dalam skripsi ini, kombinasi semua kombinasi urutan source domain menggunakan dataset penelitian Osmardifa disimulasikan untuk mengukur kinerja model menggunakan urutan pembelajaran yang berbeda dari simulasi yang dijalankan Osmardifa. Hasil simulasi urutan source domain lainnya menggunakan metode BERT menunjukkan banyak kombinasi urutan source domain yang menghasilkan kinerja lebih baik dibandingkan penelitian sebelumnya. Didapat bahwa urutan pembelajaran Capres – Jenius – Shopback – Ecom- Grab menghasilkan akurasi tertinggi 82,49% untuk retain of knowledge bagi source domain yang menggunakan dataset Capres sebagai Source Domain 1 dan urutan Capres – Jenius – Grab – Ecom – Shopback menghasilkan akurasi tertinggi 91,32% untuk transfer of knowledge. Hasil ini menunjukkan kenaikan sebesar 1,53% dan 1,72% dibandingkan simulasi awal yang dilakukan oleh Osmardifa. Analisis lanjut dilaksanakan untuk melihat apakah ada pola atau alasan yang dapat menjelaskan perbedaan kinerja pada model ketika urutan source domain digantikan akan tetapi tidak ditemukan pola atau atau alasan tersebut tidak ditemukan pada penelitian.

Technological advancements have given the public more of an opportunity to share opinions and personal evaluations within public spaces through social media and other domains on the internet.This phenomenon sparked an interest to develop a field of study under machine learning called opinion mining which specializes in analyzing sentiments found within texts. Generally, machine learning models have one domain or dataset which is used to develop the model, however with further developments a lifelong learning was developed which aims to develop models through continual learning with multiple domains or datasets. In 2022, Osmardifa underwent a study to compare the results of the Bidirectional Encoding Representations from Transfomers (BERT) model with the Convolutional Neural Network (CNN) model and the Long Short-Term Memory (LSTM) model when all of the above are used for lifelong learning. However, the comparison that was used within the study only used one combination of the sequence of source domains available using 5 source domains when there are in fact 120 possible sequences of source domains when using 5 source domains. Therefore, this study aims to further analyze the accuracy of the model in Osmardifa’s research when tested and trained using the other 120 possible learning orders of the model. Further simulations on the previously unused sequences using the BERT model showed better results than the sequence of source domains that was used in previous studies. The Capres – Jenius – Shopback – Ecom- Grab sequence showed the best resulting accuracy for the retain of knowledge tests which used the Capres dataset as the first source domain (Source Domain 1), said sequence of source domains had a final accuracy of 82.49% which is a 1.53% increase compared to previous results. The transfer of knowledge tests also showed that the Capres – Jenius – Grab – Ecom – Shopback sequence gave the best overall results with a final accuracy of 91.32% which is an increase of 1.72% compared to the previous study. Further analysis on the results of the simulations were done to check whether or not there was an underlying pattern or reason for this difference in accuracy, however no conclusive pattern or reasons were found."
Depok: Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Indonesia, 2023
S-pdf
UI - Skripsi Membership  Universitas Indonesia Library
cover
Zaid Abdurrahman
"Kemajuan teknologi memicu pertumbuhan industri teknologi dan mendorong masyarakat untuk menggunakan smartphone, terutama untuk berkomunikasi di media sosial. Media sosial merupakan tempat yang efektif untuk mencari berbagai informasi. Oleh karena itu, media sosial menyimpan banyak data, terutama data tekstual. Data tersebut muncul dari para pengguna yang jumlahnya meningkat pesat. Data tekstual bisa digunakan untuk analisis sentimen. Skripsi ini membahas analisis sentimen untuk melihat kecenderungan suatu informasi dari penulisnya. Analisis sentimen mengklasifikasikan data tekstual menjadi kelas sentimen positif dan negatif. CNN merupakan salah satu algoritma deep learning yang dapat mengklasifikasi data tekstual. Model dari algoritma CNN menunjukkan hasil yang cukup baik dalam mengkalsifikasi permasalahan analisis sentimen dengan bantuan lifelong learning. Lifelong learning merupakan machine learning yang menyerupai proses belajar pada otak manusia. Proses yang dijalankan yaitu dengan memanfaatkan hasil pembelajaran dari masa lalu untuk membantu pembelajaran pada masa depan. 4 dataset dengan domain yang berbeda, dijalankan menggunakan model CNN pada proses Lifelong learning dan menghasilkan akurasi yang meningkat, seiring dengan penambahan dataset pada proses training.

Technological advances are fueling the growth of the technology industry and encouraging people to use smartphones, especially for surfing on social media. Social media is an effective tool to find information. Therefore, social media stores a lot of data, especially textual data. The data came from users whose numbers had increased rapidly. Textual data can be used for sentiment analysis. Sentiment analysis is conducted in this study to obtain the tendency of the authors about an article. Sentiment analysis classifies textual data into a class of positive and negative sentiments. CNN is one of the deep learning algorithms that can classify textual data into positive, negative and natural classes. The model of the CNN algorithm shows good results in classifying the problem of sentiment analysis with the help of lifelong learning. Lifelong learning is a machine learning that resembles the learning process in the human brain. The process that is carried out is by utilizing learning outcomes from the past to help learning in the future. 4 datasets with different domains had ran using the CNN model in the Lifelong learning process, and produced increased accuracy along with the addition of datasets in the training process."
Depok: Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Indonesia, 2019
S-pdf
UI - Skripsi Membership  Universitas Indonesia Library
cover
Herry Susanto
"ABSTRAK
Di Indonesia, salah satu penyebab tingginya biaya BBM adalah adanya tindak pencurian
dan penyelewengan BBM yang sering kali terjadi di tengah lautan. Hal ini bisa terjadi
karena pada saat di tengah lautan, segala kegiatan kapal tersebut tidak bisa dipantau oleh
pusat operasional manajemen kapal. Selain upaya hukum, upaya pengawasan kapal
melalui teknologi terbaru juga terus dilakukan, salah satunya adalah teknologi Vessel
Monitoring System (VMS) berbasis Machine to machine (M2M). Perkembangan
teknologi VMS dan telemetri telah memungkinkan pengawasan kondisi mesin dan
pemakaian BBM kapal yang sedang berlayar secara online dan real time. Dengan
menambah perangkat pengukuran pemakaian bahan bakar tersebut, diharapkan
meningkatkan kecepatan koordinasi dan penanganan di lapangan saat terjadi
ketidakwajaran pemakaian BBM. Kecepatan dalam mengetahui adanya ketidakwajaran
ini sangat penting, karena proses pencurian minyak sering kali dilakukan dalam waktu
singkat. Pencurian minyak dengan modus ilegal tapping di darat hanya memerlukan
waktu 15 menit untuk 2000 liter (2 ton) BBM, sementara di laut diperlukan sekitar 5 jam
untuk memindahkan 12 ton BBM, atau sekitar 2.4 ton per jam untuk sebuah kapal saja.
Masalahnya untuk mengetahui ketidakwajaran tersebut masih tergantung pada analisa
tenaga ahli yang memerlukan waktu yang lama untuk melakukan analisa berbagai
parameter telemetri yang ada. Berdasarkan kondisi di atas, penelitian ini melakukan
analisis statistik terhadap data telemetri terutama data pergerakan kapal dan aktivitas
mesin untuk menentukan koefisien pergerakan kapal, lalu merancang sistem
pengklasifikasi kewajaran pemakaian BBM dengan metode Naive Bayes dan Logistic
Regression. Metode ini dipilih karena bisa memberikan hasil yang baik untuk prediksi
data-­data numerik maupun diskrit. Hasil penelitian ini menunjukkan bahwa data telemetri
dari sistem VMS dapat digunakan untuk mendeteksi adanya ketidakwajaran pemakaian
BBM. Untuk kebutuhan klasifikasi kewajaran pemakaian BBM pada data telemetri kapal,
algoritma pengklasifikasi Naive Bayes memiliki akurasi hingga 92% pada data sampel
dan Logistic Regression mampu mendeteksi dengan akurasi hingga 96% pada data
sampel.

ABSTRACT
In Indonesia, one of the causes of high fuel costs is the occurrence of theft and misuse of
fuel which often occurs in the middle of the ocean. This can happen because when in the
middle of the ocean, all the activities of the ship cannot be monitored by the ship
management operational center. In addition to legal efforts, efforts to monitor ships
through the latest technology are also being carried out, one of which is the Machine to
Machine (M2M) Vessel Monitoring System (VMS) technology. The development of
VMS and telemetry technology has enabled monitoring of engine conditions and fuel
consumption of ships that are sailing online and real time. By adding the fuel consumption
measurement device, it is expected to increase the speed of coordination and handling in
the field when there is an irregularity in the use of fuel. Speed in knowing the existence
of this irregularity is very important, because the process of oil theft is often done in a
short time. Theft of oil by illegal tapping on land only takes 15 minutes for 2000 liters (2
tons) of fuel, while at sea it takes around 5 hours to move 12 tons of fuel, or around 2.4
tons per hour for a ship. The problem is to find out the irregularities that still depend on
the analysis of experts who need a long time to analyze various parameters of existing
telemetry. Based on the above conditions, this study conducted a statistical analysis of
telemetry data, especially ship movement data and machine activity to determine the
coefficient of ship movements, then designed the fuel usage irregularity classification
system with the Naive Bayes and Logistics Regression. This method was chosen because
it can provide good results for predicting numerical and discrete data. The results of this
study indicate that telemetry data from the VMS system can be used to detect any
irregularities in using BBM. For the needs of the fairness classification of BBM usage on
ship telemetry data, the Naive Bayes classification algorithm has an accuracy of up to
92% in sample data and Logistic Regression is able to detect with accuracy up to 96% in
sample data."
2019
T53091
UI - Tesis Membership  Universitas Indonesia Library
cover
Aji Bimantoro
"Tanaman padi merupakan salah satu tanaman pangan terpenting di dunia terutama di negara-negara bagian Southeast Asia. Jumlah penduduk di dunia pun semakin meningkat setiap tahunnya sehingga kebutuhan akan makanan pokok seperti beras juga akan semakin meningkat. Namun karena adanya serangan hama dan penyakit pada tanaman padi menyebabkan kualitas dan kuantitas pada tanaman padi menurun sehingga terjadi kerugian besar dalam produksi beras. Untuk mengatasi masalah tersebut, pendeteksian penyakit pada tanaman padi menjadi sangat penting karena dapat mencegah terjadinya penurunan produksi beras. Oleh karena ini, pemrosesan data citra dan machine learning bisa menjadi salah satu cara untuk membantu mempercepat diagnosis penyakit pada tanaman padi. Pada penelitian ini, penulis menggunakan pendekatan deep learning yaitu metode Convolutional Neural Network (CNN) dengan arsitektur Xception untuk mengklasifikasi penyakit pada tanaman padi menggunakan citra daun. Data citra daun tanaman padi yang digunakan dalam penelitian ini adalah Rice Leaf Disease Image Samples yang diambil dari online database mendeley yang berisi 5932 data citra yang terdiri dari empat jenis penyakit daun padi yaitu penyakit hawar daun (Bacterial leaf blight), penyakit blas (Blast), penyakit bercak daun cokelat (brown spot), dan penyakit Tungro. Penulis melakukan tahap preprocessing sepeti crop dan resize agar ukuran citra sesuai dengan input pada model. Selanjutnya, Model akan dibangun melalui data tersebut, yang dilatih menggunakan metode CNN dengan arsitektur Xception. Data di split dengan perbandingan data latih dan data uji 70:30 dan 80:20. Kinerja model dievaluasi dengan nilai accuracy, recall, precision, dan running time. Rata-rata Accuracy, recall, dan precision yang dilakukan dalam 5 kali percobaan didapatkan pada split data 70:30 adalah masing-masing 99.708%, 99.707 %, dan 99.728% dan pada split data 80:20 masingmasing 99,662%, 99,688%, dan 99,687%. Running time yang didapatkan pada split data 70:30 adalah 43 menit dan pada split data 80:20 adalah 49 menit.

Rice is one of the most important food crops in the world, especially in Southeast Asian countries. The world's population is increasing every year so that the need for staple foods such as rice will also increase. However, due to pest and disease attacks on rice plants, the quality and quantity of rice plants decreases, resulting in huge losses in rice production. To overcome this problem, disease detection in rice plants is very important because it can prevent a decrease in rice production. For this reason, looking at image data and machine learning can be one way to help encourage disease diagnosis in rice plants. In this study, the author uses a deep learning approach, namely the Convolutional Neural Network (CNN) method with Xception architecture to classify diseases in rice plants using leaf imagery. The rice leaf image data used in this study is the Rice Leaf Disease Image Sample taken from the online mendeley database which contains 5932 image data consisting of four types of rice leaf disease, namely bacterial leaf blight and blast disease. , brown leaf spot disease (brown spot), and Tungro disease. The author performs preprocessing stages such as cropping and resizing so that the image size matches the input in the model. Furthermore, the model that will be built through the data uses the CNN method with the Xception architecture. The data is split with a comparison of training data and test data of 70:30 and 80:20. Value Performance Model with values of accuracy, recall, precision, and running time. The average accuracy, recall, and precision carried out in 5 trials at the 70:30 data split were 99.708%, 99.707%, and 99.728%, respectively, and in the 80:20 data split they were 99.662%, 99.688%, respectively, and 99.687%. The running time obtained in the 70:30 data split is 43 minutes and the 80:20 data split is 49 minutes."
Depok: Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Indonesia, 2022
S-pdf
UI - Skripsi Membership  Universitas Indonesia Library
cover
Scales, Peter, 1949-
"An essential book linked to the LLUK Standards for teachers, trainers &​ tutors: a practical introduction to teaching &​ learning.
This popular introductory textbook is ideal for anyone working or training to work in the lifelong learning sector. The new edition has been comprehensively revised to reflect recent developments in the sector and current research in learning and teaching. The book covers key topics such as reflective teaching, communication, learning theories, and assessment for learning. In addition there are new chapters on: Behaviour for learning; A curriculum for inclusive learning; The lifelong learning sector and Functional skills. This edition also includes more student journal extracts, case studies and developmental activities. Common elements of good practice in teaching and learning spanning the lifelong learning, further education and skills sector and are fully explored so that you will: Gain a thorough understanding of learners and their needs Understand the importance of effective communication Appreciate the role of reflective practice and continuing professional development Achieve a good grasp of theory and practice including methods of active learning and assessment for learning Teaching in the Lifelong Learning Sector is essential reading for those teaching or training to teach in further and higher education, adult and community learning, and work-based learning.
This popular introductory textbook is ideal for anyone working or training to work in the lifelong learning sector. This new edition has been comprehensively revised to reflect recent developments in the sector and current research in learning and teaching."
New York: Open University Press, 2013
371.102 SCA t
Buku Teks  Universitas Indonesia Library
cover
Delphin, Patricia
Virgnia Reston publishing 1983 , 1983
610.730 7 Dol c
Buku Teks SO  Universitas Indonesia Library
cover
Nadya Asanul Husna
"Inhibitor DPP-4 adalah pendekatan baru yang menjanjikan untuk pengobatan diabetes tipe-2 dengan risiko rendah hipoglikemia. Pemodelan hubungan kuantitatif struktur-aktivitas (QSAR) adalah pemodelan yang digunakan untuk menyaring basis data besar suatu senyawa untuk menentukan sifat biologis molekul kimia berdasarkan struktur kimianya. Pada tesis ini pemodelan QSAR yang digunakan adalah QSAR klasifikasi dan QSAR regresi. Sebelum membuat model QSAR akan melakukan esktraksi ciri pada struktur molekul (SMILES). Hasil ekstraksi ciri tersebut kemudian akan digunakan sebagai masukan untuk metode rotation forest kasus klasifikasi dan kasus regresi. Model QSAR klasifikasi akan memprediksi molekul aktif dan tidak aktif pada inhibitor DPP-IV. Sedangkan model QSAR regresi akan memprediksi nilai aktivitas IC50 inhibitor DPP-IV. Pada penelitian ini untuk kasus klasifikasi dan regresi juga membandingkan performa model rotation forest menggunakan matriks rotasi PCA dengan rotation forest menggunakan matriks rotasi Sparse PCA.
Hasil penelitian ini menunjukkan bahwa model QSAR regresi menggunakan rotation forest dengan matriks rotasi PCA (RFR(PCA)) memperoleh koefisien korelasi kuadrat 29.2% dengan RMSE 45%. Sementara itu, menggunakan rotation forest dengan matriks rotasi Sparse PCA (RFR(SPCA)) memperoleh koefisien korelasi kuadrat 27.1% dengan RMSE 45.6%. Pada QSAR klasifikasi persentase banyaknya molekul yang aktif sangat besar dibandingkan yang molekul tidak aktif, hal ini dapat menyebabkan nilai evaluasi berbeda. SMOTE (Synthetic Minority Oversampling Technique) merupakan salah satu metode untuk menangani data tidak seimbang tersebut dengan cara membangkitkan data buatan. Hasil penelitian ini menunjukkan bahwa model QSAR klasifikasi menggunakan rotation forest dengan matriks rotasi PCA (RFC(PCA)) memperoleh performa tertinggi dalam memprediksi molekul aktif dan tidak aktif, yaitu nilai MCC 77.7% dengan nilai akurasi sebesar 89%, sensitivitas 89.6%, dan spesifisitas 88.1%. Sementara itu, model QSAR klasifikasi menggunakan rotation forest dengan matriks rotasi SPCA (RFC(SPCA)) memperoleh performa tertinggi, yaitu nilai MCC 80.9% dengan nilai akurasi sebesar 90.5%, sensitivitas 90.8%, dan spesifisitas 90.2%.

DPP-4 inhibitors are a new approach for the treatment of type 2 diabetes with a low risk of hypoglycemia. The Quantitative Structure-Activity Relationship (QSAR) model is a model used to filter large databases of compounds to determine the biological properties of chemical molecules based on their chemical structure. The QSAR modeling that is used in this research is QSAR classification and QSAR regression. Before creating the model, QSAR will perform feature extraction on the molecular structure (SMILES). The results of the feature extraction will be used as inputs for the rotation forest method of the classification and regression cases. The QSAR classification model predicts active and inactive molecules in DPP-IV inhibitors, while the regression QSAR model predicts the value of IC50 DPP-IV inhibitor activity. In this study, the classification and regression cases are also comparing the performances between the rotation forest model using the PCA rotation matrix and the rotation forest model using the Sparse PCA rotation matrix. 
The results of this study indicate that the QSAR regression model using rotation forest with the rotation matrix PCA (RFR (PCA)) obtained a squared correlation coefficient of 29.2% with RMSE 45%. Meanwhile, using rotation forest regression with the Sparse PCA (RFR (SPCA)) rotation matrix obtained a quadratic correlation coefficient of 27.1% with RMSE 45.6%. In the QSAR classification, the percentage of active molecules is very large compared to inactive molecules, this can cause different evaluation values. SMOTE (Synthetic Minority Oversampling Technique) is one method for handling such unbalanced data by generating artificial data. The results of this study indicate that the classification QSAR model using rotation forest classification with PCA (RFC (PCA)) rotation matrix obtained the highest performance in predicting active and inactive molecules as follows: MCC value of 77.7% with an accuracy value of 89%, sensitivity value of 89.6% and specificity value of 88.1%. Meanwhile, the QSAR classification model using rotation forest classification with the SPCA rotation matrix (RFC (SPCA)) obtained the highest performance as follows: MCC value of 80.9% with an accuracy value of 90.5%, sensitivity value of 90.8%, and specificity value of 90.2%.
"
Depok: Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Indonesia, 2020
T-pdf
UI - Tesis Membership  Universitas Indonesia Library
cover
"The second edition of the International Handbook of Lifelong Learning is extensive, innovative, and international in scope, remit and vision, inviting its readers to engage in a critical re-appraisal of the theme of “lifelong learning”. It is a thorough-going, rigorous and scholarly work, with profound and wide-ranging implications for the future of educating institutions and agencies of all kinds in the conception, planning and delivery of lifelong learning initiatives. Lifelong learning requires a wholly new philosophy of learning, education and training, one that aims to facilitate a coherent set of links and pathways between work, school and education, and recognises the necessity for government to give incentives to industry and their employees so they can truly “invest” in lifelong learning. It is also a concept that is premised on the understanding of a learning society in which everyone, independent of race, creed or gender, is entitled to quality learning that is truly excellent.
This book recognises the need for profound changes in education and for goals that are critically important to education, economic advancement, and social involvement. To those concerned about the future of our society, our economy and educational provision, this book provides a richly illuminating basis for powerful debate. Drawing extensively on policy analyses, conceptual thinking and examples of informed and world-standard practice in lifelong learning endeavours in the field, both editors and authors seek to focus readers' attention on the many issues and decisions that must be addressed if lifelong learning is to become a reality for us all.
"
Dordrecht: Springer Science, 2012
e20426598
eBooks  Universitas Indonesia Library
cover
Livingstone, D.W.
"This book presents some of the most trenchant critical analyses of the widespread claims for the recent emergence of a knowledge economy and the attendant need for greater lifelong learning. The book contains two sections. First, general critiques of the limits of current notions of a knowledge economy and required adult learning, in terms of historical comparisons, socio-political construction and current empirical evidence. Secondly, specific challenges to presumed relations between work requirements and learning through case studies in diverse current workplaces that document richer learning processes than knowledge economy advocates intimate. "
Rotterdam : Sense, 2012
e20401067
eBooks  Universitas Indonesia Library
cover
Gita Kartika Suriah
"Analisis sentimen merupakan suatu proses untuk menentukan sikap atau sentimen dari penulis mengenai hal tertentu. Proses pengelompokan sentimen secara manual membutuhkan waktu cukup lama, sehingga diusulkan untuk menggunakan machine learning. Pada penelitian ini, model machine learning yang digunakan merupakan model CNN-BiLSTM (Convolutional Neural Network - Bidirectional Long Short-Term Memory) dan BiLSTM-CNN (Bidirectional Long Short-Term Memory - Convolutional Neural Network) yang menghasilkan kinerja yang lebih baik dibandingkan model CNN dan BiLSTM pada permasalahan analisis sentimen. Supaya model dapat belajar secara berkelanjutan dari beberapa domain data, model tersebut juga diimplementasikan lifelong learning. Hasilnya, model CNN-BiLSTM menunjukkan kinerja transfer of knowledge yang lebih baik dibandingkan oleh model BiLSTM-CNN maupun model dasarnya. Di sisi lain, model BiLSTM-CNN menunjukkan kinerja yang lebih buruk dibandingkan model dasarnya. Sedangkan, hasil loss of knowledge menunjukkan bahwa kinerja model CNN- BiLSTM lebih buruk dari BiLSTM-CNN. Selain itu, kedua model gabungan tersebut menunjukkan kinerja yang lebih baik dibandingkan model CNN, tetapi lebih buruk dibandingkan model BiLSTM. Untuk pengembangan lebih lanjut, diimplementasikan pula lifelong learning dengan pembaruan vocabulary. Dengan implementasi tersebut, model mampu mempelajari vocabulary dari domain data 2, 3, 4, dan 5. Pembaruan vocabulary ternyata meningkatkan kinerja model pada transfer of knowledge dan loss of knowledge.

Sentiment analysis is a process to determine the attitude or sentiment of the author regarding certain matters. The process of classifying sentiments manually takes a long time, so it is proposed to use machine learning. In this study, the machine learning model used is the CNN-BiLSTM (Convolutional Neural Network - Bidirectional Long Short-Term Memory) and BiLSTM-CNN (Bidirectional Long Short-Term Memory - Convolutional Neural Network) models which produce better performance than the CNN and BiLSTM models on the problem of sentiment analysis. In order for the model to learn continuously from several data domains, the model is also implemented lifelong learning. As a result, the CNN-BiLSTM model shows better transfer of knowledge performance compared to the BiLSTM-CNN model and its base model. On the other hand, the BiLSTM-CNN model shows a worse performance than its base model. Meanwhile, the results of loss of knowledge show that the performance of the CNN-BiLSTM model is worse than the BiLSTM-CNN model. In addition, the two combined models show better performance than the CNN model, but worse than the BiLSTM model. For further development, lifelong learning is also implemented with an update to vocabulary. With this implementation, the model is able to learn vocabulary from data domain 2, 3, 4, and 5. In fact, the vocabulary update has an effect in increasing the performances of transfer of knowledge and loss of knowledge.
"
Depok: Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Indonesia, 2021
S-pdf
UI - Skripsi Membership  Universitas Indonesia Library
<<   1 2 3 4 5 6 7 8 9 10   >>