پديد آورنده :
تاكي، بهنام
عنوان :
تشخيص ناهنجاري در سري‌هاي زماني به كمك شبكه‌هاي عصبي بازگشتي
مقطع تحصيلي :
كارشناسي ارشد
گرايش تحصيلي :
هوش مصنوعي و رباتيكز
محل تحصيل :
اصفهان : دانشگاه صنعتي اصفهان
صفحه شمار :
يازده، 69ص. :مصور، جدول، نمودار
استاد راهنما :
عبدالرضا ميرزايي
استاد مشاور :
مهران صفاياني
توصيفگر ها :
تشخيص ناهنجاري , سري‌هاي زماني , شبكه‌هاي حافظه كوتاه‌مدت ماندگار , يادگيري ژرف , يادگيري بدون‌نظارت
استاد داور :
جواد عسگري، الهام محمودزاده
تاريخ ورود اطلاعات :
1398/06/13
رشته تحصيلي :
مهندسي كامپيوتر
دانشكده :
مهندسي برق و كامپيوتر
تاريخ ويرايش اطلاعات :
1398/06/13
چكيده انگليسي :
Anomaly Detection in Time series by means of Recurrent Neural Networks Behnam Taki b taki@ec iut ac ir 30 June 2019 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 IranDegree Master of Science Language FarsiSupervisors Abdolreza Mirzaei Assist Prof mirzaei@cc iut ac ir Mehran Safayani Assist Prof safayani@cc iut ac ir AbstractNowadays with the rapid growth of data in variation and volume highlights are on autoanalysis methods autoanalysis is a technique for extracting information from data one of the most popular applications forautoanalysis is anomaly detection as a basic machine learning problem the purpose of anomaly detection is toprovide a diagnostic understanding of the abnormal data generative process by assuming that the probability orlikelihood of the process generating the normal data is as large as possible the time series anomaly detectionscenarios arise in the context of many applications such as medical data sensor data or network intrusion in temporal data the data are not expected to change abruptly unless there are abnormal processes at work inthis thesis a new anomaly detection method on time series with an unsupervised approach and by the use ofLSTM explored the proposed method divided into 2 main parts multi step predictions and modes declaration preserving serially dependent values in time series by LSTM made us a powerful tool in both parts we calculatederror predictions from the first part and fit them to three normal distributions in the second part the likelihoodof each prediction errors suggests as an anomaly score the results and experiments show that our new approachhas a good performance even better than its related works KeywordsAnomaly detection Time series Long short term memory Deep learning Unsupervised learning
استاد راهنما :
عبدالرضا ميرزايي
استاد مشاور :
مهران صفاياني
استاد داور :
جواد عسگري، الهام محمودزاده
