پديد آورنده :
رضايي، محمدرضا
عنوان :
تخمين موقعيت و سرعت موش آزمايشگاهي در يك مارپيچ با كمك پردازش فعاليت اسپايكي جمعيت نوروني با يادگيري عميق
مقطع تحصيلي :
كارشناسي ارشد
محل تحصيل :
اصفهان : دانشگاه صنعتي اصفهان
صفحه شمار :
چهارده، 83ص.: مصور (رنگي)، جدول، نمودار
استاد راهنما :
بهزاد نظري، سعيد صدري
توصيفگر ها :
شبكه‌هاي عصبي پسماند , شبكه‌هاي عصبي پيچشي , مدل‌سازي آماري , فيلتر فرايند نقطه‌اي , پردازش سيگنال‌هاي بيولوژيكي
استاد داور :
محمدرضا احمدزاده
تاريخ ورود اطلاعات :
1398/03/27
رشته تحصيلي :
برق و كامپيوتر
دانشكده :
مهندسي برق و كامپيوتر
تاريخ ويرايش اطلاعات :
1398/03/27
چكيده انگليسي :
Deep LSTM neural network for position estimation of rat using spiking activity of neural population Mohammad Reza Rezaei mreza rezaei@ec iut ac ir May 20 2019 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Drs Behzad Nazari Saeid Sadri and Ali yousefi nazari@cc iut ac ir Abstract Generally in neural systems of animals information about behavioural variables such as sensory signals or motor actionsare carried by the activity of populations of neurons Signal decoding of neural system helps researchers to understand brain sfunctions Research about hippocampus area of rat brain shows that it is possible to decode position from spiking activityof this area Recently with the emergence of deep learning techniques deep neural networks DNNs provide powerfulmodeling capabilities and achieve state of the art results Long short term memory LSTM a type of deep neural network can capture long range dependencies and nonlinear dynamics and is widely used in modeling complex dynamical signalslike speech In this research we propose long LSTM network topologies for decoding 2D movement trajectory of a rat usingthe neural activities recorded from an ensemble of hippocampal place cells Despite of wide utilization of DNNs reliabilityof result in these modeling approach is not completely understood To analyze the behavior and accuracy of the networks wecompared the performance of these topologies with point process filter solution which is used widely in these experiments The result of the first proposed topology in this thesis even though it has sufficient decoding accuracy shows undesiredbehaviors which are unnatural in rat s movement like jumps The second proposed topology restricts output to produceundesired jumps by embedding information of the maze shape Finally in the third proposed topology adding informationof rat velocity to the model improves dynamics of rat s movement in decoding Since the rat s position distribution is notuniform in training session the cost function is weighted by absolute value of the velocity We measured least absoluteerror LAE and root mean square error RMSE to compare the performance of these methods LAE metric for LSTMand point process filter are 8 7 and 6 8 respectively RMSE metric is 10 3 and 8 86 for LSTM and point process filter respectively We showed that the LSTM topologies and point process filter provide comparable accuracy in estimating theposition In addition both the LSTM model and the point process model can encode the receptive field for each place cell The LSTM runs 16 times faster than the point process filter in this research providing a strong advantage in computationalefficiency Key Words Deep neural network biomedical signal processing LSTM neural network Point processfilter Recurrent neural network Neural decoding Neural data analysis
استاد راهنما :
بهزاد نظري، سعيد صدري
استاد داور :
محمدرضا احمدزاده
