Courses

3 different courses will be held on October 27th, 2018,

- Participation to each course should be minimum 5 people.

- Course price is valid only for one Course. Delegates should participate only one course at one time.

- For registatration Courses please contact with Organizaation Secratariat.

 

Course Price : 150 .-TL

 

Services included in Course price ;

- Participation to Courses, tea-coffee services according to Course Programme, Participation Certificate and VAT

 

COURSE 1 :  Basic R Programming / Prof. Dr. Erdem KARABULUT – Doç. Dr. Beyza DOĞANAY ERDOĞAN

 

COURSE 2 :  Monte Carlo Simulations / Doç. Dr. Ertuğrul ÇOLAK

Monte Carlo Simulations

Monte Carlo relates to or involves "the use of random sampling techniques and often the use of computer simulation to obtain approximate solutions to mathematical or physical problems especially in terms of a range of values each of which has a calculated probability of being the solution". This definition provides a concise and accurate description for Monte Carlo Simulations. Monte Carlo simulation has become an important and popular research tool used by quantitative researchers in biostatistics. The Monte Carlo method provides approximate solutions to a variety of statistical problems by performing statistical sampling experiments via computer. Monte Carlo simulation offers researchers an alternative to the theoretical approach; this is important because many situations exist in which implementing a theoretical approach is difficult and finding an exact solution is even more difficult. In addition, computing power has become increasingly less expensive and computers are more widely available than ever before.

Learning outcome and competences

After completion of the course, it is expected that the participant has an overview of a broad spectrum of methods used in Monte Carlo simulation, in particular random variate generation from a specified distribution, output analysis, variance-reduction methods, simulation of stochastic processes. In addition, the participant has an understanding of the scope and limitations of the Monte Carlo method, and a working skill in programming in SAS University Edition.

Software

SAS University Edition will be used during the course. SAS University Edition is a free SAS® software for academic, noncommercial use. The participant can download the program from https://www.sas.com/en_us/software/university-edition.html.  

Course Details

  • Course Introduction

  • Basic Procedures for Monte Carlo Simulation

  • Generating Data in Monte Carlo Studies

  • Automating Monte Carlo Simulations

  • Conducting Monte Carlo Studies That Involve Univariate Statistical Techniques

  • Conducting Monte Carlo Studies for Multivariate Statistical Techniques

  • Number of Replications Required in Monte Carlo Simulation Studies

 

 

COURSE 3 : How to do Biological Data Analysis with Deep Learning? / Dr. Erdal COŞGUN

 Deep Learning is a sub-class of machine learning algorithms that learn from large and multi-layered aggregated agglomerations and can analyze these processors in a wide variety of examples. The design of this layered structure of algorithms, called artificial neural networks, is inspired by the biological neural network used by the human brain. While it is expected that a standard machine learning model should make a correct prediction (feeding/training more data), the Deep Learning model can be learned on its own. It is similar to how a human brain will perceive a problem, think about it, and then produce a conclusion. Developments in the design of Deep Learning algorithms occur at the same time as rapid development, such as rapid and large-scale information storage capacity, high computing power and paralel processing.

The most well-known deep learning algorithms are Deep Neural Networks, which is obtained from many layers of linear and non-linear processing units (hence the use of "deep" terminology) and trained with large amounts of training datasets. Another Deep Learning algorithm is the Random Decision Forest (RDF). They are also built from many layers, but are built from RDF decision trees instead of neurons and produce statistical averages (median or mean) of estimates of individual trees. Within the scope of this training, the biostatistics specialists / candidates, who are inevitably part of the health research, will have the opportunity to discuss the theoretical details of the Deep Learning approaches / libraries which will affect the researhes in next decade. This analysis group, which is difficult to implement with personal computers, will be able to train with real data sets.

 

Agenda

Opening Session

How can we define big data in medical researches?

Key differences between Machine Learning and Deep Learning 

Understanding how can we perfom the Deep Belief Networks & Deep Autoencoder?

How can we use Graphics processing unit (GPU) infrastuructures for Biostatistics

Keras Applications: Defining genetically similar patients with Deep Autoencoder

Developing Deep Learning prediction models with Jupyter notebook

Q&A and Closing

 

Sample Data Sets

  1. VCF data sets- TBD

  2. Sample classification data set: https://cntk.ai/pythondocs/CNTK_101_LogisticRegression.html

 

Software Resources:

  1. https://www.anaconda.com/download/

  2. https://www.python.org/downloads/

  3. https://github.com/Microsoft/CNTK

 

References:

  1. https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/DeepLearning-NowPublishing-Vol7-SIG-039.pdf

  2. http://deeplearning.net/reading-list/ 

  3. http://deeplearning.net/tutorial/

  4. http://ufldl.stanford.edu/wiki/index.php/UFLDL_Tutorial   

  5. http://ufldl.stanford.edu/wiki/index.php/UFLDL_Recommended_Readings

  6. https://github.com/Microsoft/CNTK/

  7. https://github.com/Microsoft/CNTK/blob/master/PretrainedModels/Image.md

  8. https://cntk.ai/pythondocs/tutorials.html