Hasil Pencarian  ::  Simpan CSV :: Kembali

Abstrak :
ABSTRAK

Dionaea adalah honeypot yang memiliki tujuan utama mendapatkan salinan dari malware. Setelah mendapatkan salinannya, proses knowledge discovery dilakukan untuk mendapatkan informasi dari database Dionaea. Dua alasan utama penggunaan knowledge discovery adalah data terlalu banyak namun informasinya sedikit, dan untuk mengekstrak informasi yang berguna dan menafsirkannya. Proses knowledge discovery memiliki beberapa fase, yaitu pembersihan data, seleksi data, transformasi data, prekalkulasi, data mining, evaluasi pola, dan penyajian informasi. Proses data mining menggunakan variasi algoritma DBSCAN, yaitu multidensity DBSCAN. Analisis dibagi menjadi dua, yaitu analisis cluster dan dataset. Analisis cluster menjelaskan hubungan antara lokasi negara penyerang berdasarkan daerah Internet Registry-nya dan persentase deteksi malware berdasarkan beberapa vendor antivirus. Dari analisis dataset, didapatkan informasi bahwa malware yang paling sering digunakan penyerang atau tren malware, berjenis Downadup, yaitu sebesar 71.1%. Negara yang paling sering menyerang adalah Rusia dan beberapa negara Eropa. Sebagai pembanding, laporan tahunan yang dipublikasi Microsoft, ENISA, dan F-Secure pada akhir 2014 menunjukkan tren malware yang sama, yaitu berjenis Downadup.

---

ABSTRACT

The main purpose of implementation of Dionaea is to get copy of malwares. After that, knowledge discovery is applied to get information from Dionaea?s database. Two main reasons to use data mining method are data is too large but only contain few informations, and to extract useful informations and interpret them. Knowledge discovery process have several steps, they are data cleaning, data selection, data transformation, precalculation, data mining, pattern evaluation, and knowledge representation. Data mining process uses multidensity DBSCAN. There are two main sections of analysis, cluster analysis and dataset analysis. Cluster analysis show the relation between attackers? country location which is based on their Regional Internet Registry and malware detection rate from several antivirus vendor. Dataset analysis shows the most frequent country whose attacker is Conficker variant, 71.1% of all dataset is Conficker worm incident and the mode of attacker country is Russia and severals Europe countries. This outputs show similarity about threat landscape and malware in Asia, compared to annual report by Microsoft, Enisa, and F-Secure which was published at the end of 2014, which stated Downadup as most frequent malware.

Abstrak :
Learning in non-stationary environments : methods and applications offers a wide-ranging, comprehensive review of recent developments and important methodologies in the field. The coverage focuses on dynamic learning in unsupervised problems, dynamic learning in supervised classification and dynamic learning in supervised regression problems. A later section is dedicated to applications in which dynamic learning methods serve as keystones for achieving models with high accuracy. Rather than rely on a mathematical theorem/proof style, the editors highlight numerous figures, tables, examples and applications, together with their explanations.

Abstrak :
This volume offers comprehensive coverage of state-of-the-art ensemble learning techniques, including the random forest skeleton tracking algorithm in the Xbox Kinect sensor, which bypasses the need for game controllers. At once a solid theoretical study and a practical guide, the volume is a windfall for researchers and practitioners alike.

Abstrak :
ABSTRAK
Balai Jaringan Informasi dan Komunikasi (BJIK) sebagai salah satu balai di Badan Pengkajian dan Penerapan Teknologi (BPPT) memiliki tugas dalam penerapan teknologi informasi dan komunikasi (TIK) untuk kepentingan pemerintah pusat, daerah, publik, komunitas ilmu pengetahuan teknologi, dan industri. Tugas tersebut diwujudkan salah satunya dengan membangun sistem informasi monitoring teknologi informasi dan komunikasi yang bernama Simontik. Kemajuan tren teknologi dan ancaman siber yang tidak dapat dihindari membutuhkan adanya penerapan data mining untuk monitoring intrusi dalam melindungi informasi penting dimana perangkat lunak anti virus dan firewall tidak cukup memberikan perlindungan penuh sesuai dengan kondisi BJIK saat ini. Sejalan dengan hal tersebut, beberapa penelitian terdahulu juga menjelaskan teknik deep learning atau deep neural network pada data mining yang telah mencapai keberhasilan jauh lebih baik di berbagai aplikasi khususnya big data sets classification karena memberikan hasil yang akurat dalam menyelesaikan permasalahan sistem monitoring intrusi. Berdasarkan hal tersebut, penelitian ini menggunakan teknik classification dengan algoritme deep learning, support vector machine, dan random forest sebagai pembanding. Penelitian ini menggunakan metodologi knowledge discovery from data (KDD) dimana data mining hanya merupakan suatu langkah penting dalam urutan prosesnya. Hasil akhir dari penelitian ini merupakan model prediksi yang dikemudian diuji dengan dataset Simontik untuk diketahui akurasinya. Hasil yang didapatkan dari penelitian ini adalah algoritme deep neural network dan random forest menghasilkan akurasi yang paling baik, yaitu sebesar 99,91% dibandingkan dengan algoritme support vector machine yang memiliki akurasi sebesar 98,11%. ABSTRACT
The Information and Communication Network Center (BJIK) as one of the centers in the Agency for the Assessment and Application of Technology (BPPT) has the task of implementing information and communication technology (ICT) for the benefit of the central, regional, public, technological and industrial science communities. One of the tasks is realized by building an information and communication technology monitoring information system called Simontik. The unavoidable progress of technological trends and cyber threats requires the application of data mining for intrusion monitoring in protecting important information where anti-virus software and firewalls do not provide full protection in accordance with current BJIK conditions. In line with this, several previous studies also explained that deep learning techniques or deep neural networks in data mining that have achieved success are far better in various applications, especially the big data sets classification because they provide accurate results in solving intrusion monitoring system problems. Based on this, this study uses classification techniques with deep learning algorithms, support vector machines, and random forest as a comparison. This study uses the knowledge discovery from data (KDD) methodology where data mining is only an important step in the sequence of the process. Result of this study is a prediction model which is then tested with the Simontik dataset to determine its accuracy. The results obtained from this study are that deep neural network and random forest algorithms produce the best accuracy, which is 99.91% compared to the support vector machine algorithm which has an accuracy of 98.11%.

Abstrak :
This book constitutes the refereed proceedings of the Second International Conference on Theory and Practice of Digital Libraries, TPDL 2012 - the successor of the ECDL (European Conference on Research and Advanced Technology for Digital Libraries) - held in Paphos, Cyprus, in September 2012. The 23 full papers, 19 short papers, 15 posters and 8 demonstrations presented in this volume were carefully reviewed and selected from 139 submissions. The papers are organized in topical sections on user behavior, mobiles and place, heritage and sustainability, preservation, linked data, analysing and enriching documents, content and metadata quality, folksonomy and ontology, information retrieval, organising collections, as well as extracting and indexing.

Abstrak :
This book features the proceedings of the Fifth International Conference on Computational Science and Technology 2018 (ICCST2018), held in Kota Kinabalu, Malaysia, on 29–30 August 2018. Of interest to practitioners and researchers, it presents exciting advances in computational techniques and solutions in this area. It also identifies emerging issues to help shape future research directions and enable industrial users to apply cutting-edge, large-scale and high-performance computational methods.