Mohsen Davarynejad, Ph.D.
Persian: محسن داوری نژاد
This website offers information on my academic and professional career. It also provides access to some of my publications, and to projects I am or have been involved in. The following is the tag-cloud of the title of my publications and it reflects my past / current interests.
Mohsen is a senior data science consultant responsible for scoping, designing and implementing analytic solutions featuring AI, Deep Learning, Machine Learning and Text Mining. He has a profound knowledge of advanced planning and scheduling problems. In particular he has more than ten years of experience in applying and innovating optimization algorithms in a wide range of domains.
His academic background has resulted in strong analytical skills and a structured approach to problem-solving. With several years of experience in many fields (Research, Healthcare and Process Industries) Mohsen can take charge of analytical roles starting from the tool design (questionnaire, interview creation) to data visualization (Spotfire and Tableau) and data mining.
He has a strong knowledge of statistical analysis and machine learning using numerous tools (Python -Numpy, Scipy, Scikit-Learn-, H20, Spark, Scala, Matlab, KNIME, SPSS, Big data tools like Hadoop, Cloud computing like Azure, Databricks and Database systems like Teradata).
Have a look at the latest NEWS around my professional activities.
Key Competencies and Strengths:
- Over 12 years of work and research experience in Machine Learning and Data Science.
- Strong data analytical and programming skills especially with Python and R.
- Architect, develop and deploy data solutions in Azure cloud and big data tools: Databricks and pySpark
- Certified Azure Data Scientist and Azure Data Engineer See credentials.
- Excellent English writing and oral presentation skills.
- Strong team-work spirit with experience of working in highly international environments for years.
- Designing systems of insight: Designing systems to help organisations become data-driven.
- OR: Designing algorithms capable of finding an optimal solution for a complex optimization problem.
- Scalable and optimized Azure cloud solution.
- Predictive analytics,
- Operations research and Optimization,
- Reinforcement Learning,
- I have applied many (including my) algorithms to scientific and engineering problems in fields as diverse as traffic network control, aviation industry, supply chain management and medical science. In the latter area, I have worked on problems of intensity-modulated radiation therapy planning.
- Numerous career-relevant extracurricular activities (teaching, conference organiser), all of which have developed both my technical and communication skills to a high standard.
- Developing android apps, using Parse.com as the Backend.
- Feel free to connect/contact if you have an interest in Data Analytics, Optimization and Entrepreneurial Ideas.
Delft University of Technology, Delft, the Netherlands 2009 - 2014
Ferdowsi University of Mashhad, Mashhad, Iran 2000.0 - 2000.0
M.Sc. in Electrical Engineering (Control)
Cognitive Computing Laboratory, Center for Applied Research on Soft Computing and Intelligent Systems, Electrical Engineering Department, Ferdowsi University of Mashhad, Iran.
Ferdowsi University of Mashhad, Mashhad, Iran 2000.4 - 2000.11
B.Sc. in Electrical Engineering (Control)
Ferdowsi University of Mashhad (FUM), Electrical Engineering Department, Mashhad, Iran.
My broad area of interest is Intelligent Data Analysis (particularly database mining techniques) and Computational Intelligence.
How to learn representations of real-world, large, various unlabeled data??
A list of my publication should be found on my website
SENIOR CONSULTANT DATA SCIENCE AND ANALYTICS
|Jun. 2016 - Present||
Involved in a number of projects including the top four being:
Anomaly detection for Predictive Asset Maintenance (PAM)
Text Data Mining and ontology learning for smart matching
Predicting LatePax (Late passengers at the gate)
Predictive analytics for aircraft arrival times
Social media data analytics
Mohsen was also involved in writing proposals for potential clients and execution of PoCs (e.g: implementing a Chabot using Microsoft bot framework using c# and LUIS.ai).
Tooling: Python (Numpy, Pandas, Scipy, Scikit-Learn), Python Flask, Python Dash, Jupyter Notebook, Git, H2O, R, Hive, Airflow, AWS, Docker, PI, Seeq, Spotfire, Power BI
Methodologies: NN, SVM, PCA, autoencoders (Diabolo network), Ensembles of Classifiers, AdaBoost, Extreme Gradient Boosting, Word2Vec, Doc2Vec, Graph analysis, Association Rules and the Apriori Algorithm
CAPGEMINI NETHERLANDS B.V.
SENIOR CONSULTANT DATA SCIENCE AND ANALYTICS
|Aug. 2015 - May 2016||
Involved in a number of projects including:
Implementing a recommendation system for an international grocery store on Hadoop Cluster using MLlib on Spark.
Implementing a recommendation system for an Academy Institution partly on IBM Watson.
Managing a team of 5 data scientists and developers for an app development project.
Lean core team member and project manager data science at Philips: Standardizing IT and business processes, VSM, facilitating workshops.
Tooling: Python (Numpy, Pandas, Scipy, Scikit-Learn), R, Hadoop ecosystem (Cloudera), Spark, Spotfire, KNIME, WEKA (an open source collection of Machine Learning algorithms for Data Mining tasks), SPSS
Methodologies: Recommendation systems, Machine learning, Evolutionary computation
|Aug. 2013 - Jul. 2015||
The TREsPASS project is a cross sector scientific project with the objective to predict,
prioritise, and prevent complex attacks
(e.g. information security threats to organisations, physical and social engineering threats) in a systematic way.
As a scientific researcher Mohsen took up the responsibility of leading a work package on
"Abstraction levels for model sharing".
Together with an international group of more than 12 people they have reviewed the state-of-the-art industry processes
and tools for sharing sensitive information between organisations and proposed a design for an standard library service,
instantiating it with the Model Pattern Library (MPL) and Attack Pattern Library (APL).
This has resulted in a deliverable for the TREsPASS project and a number of scientific publications under the name of Mohsen.
Moreover Mohsen has worked on projects such as "estimating attack frequencies based on limited attacker resources" as well as projects on real-world problems mainly in logistics, traffic control and management with the application of data analysis, data mining and machine learning.
ERASMUS MEDICAL CENTER
|Jan. 2013 - Dec. 2014||
Treatment planning for radiation therapy
inherently involves trade offs, such as between target coverage and normal tissue sparing. During his stay at the
EMC Mohsen was responsible for
developing methods that can facilitate decision making related to such trade offs by approximation of
Pareto optimal sets. He was also responsible for refining and
extending automated treatment planning with the
help of faster optimization algorithms
(iterative constrained optimizers,
interior-point optimizers and the
normal constraint methods).
After finalizing the design and validating this on Matlab, together with a team of
developers they have integrated it into the EMC in-house software for clinical use.
RESEARCH SOFTWARE ENGINEER
|Jan. 2009 - Jul. 2009||To stay competitive in the market, Trinité Automation wanted to offer advanced monitoring and control strategies to its clients. In his role as a research software engineer Mohsen was responsible for requirements analysis and screening, prototyping/implementing and testing of various data fusion algorithms and control systems. He designed and implemented a tool for estimation of missing data using Treiber-Helbing filter. He was also responsible for implementation and testing of FileProof project (for Traffic Control Center of the Province of North-Holland - the project includes 2462 sensors and 1599 actuators which are distributed in a network of 80 by 50 km) with a team of 8 developers. Completion of the testing phase is accomplished through preliminary use by actual end user.|
|Oct. 2008 - Dec. 2012||
Involved in a number of projects including:
Design, implement and evaluate fast optimization strategies for complex optimization problems: Computational complexity is still a prohibitive factor in evolutionary optimization of sufficiently large and/or complex problems. The goal of my PhD research was to design, implement and evaluate fast optimization strategies for complex optimization problems, advanced data analysis and trend prediction algorithms. The ideas and algorithms were implemented and evaluated in Matlab, R and Python. The result of his studies are presented in a number of conferences and high-ranked journals and has received considerable attention in the scientific community in the last few years.
Time series models for motorway flow forecasting: For online road traffic control at traffic control centers, there is a need for predictions of the traffic flow. For this effort, predictions are needed of the traffic inflow into the network. Mohsen was responsible for developing fast and accurate prediction algorithm for the inflows into the network. His proposed prediction algorithms are based on neural networks and support vector machines. The algorithms are designed and tested on traffic flow data of the ring road of Amsterdam. This resulted in a robust predictions of traffic demand with relatively small errors for the next 30 min in a large-scale real-time environment and shows improvement by at least 20%.
Measuring operational performance of airport traffic and transportation: Amsterdam Airport Schiphol is well known as Holland’s biggest mobility node. Within Schiphol the Traffic & Transportation department has the responsibility for the accessibility of Schiphol. The goal of the Traffic & Transportation department is to improve the accessibility for Schiphol’s customers. As a scientific researcher Mohsen was responsible for designing and building a system to measure Schiphol’s landside accessibility. Build on top of a master thesis that he has supervised, he suggested a multi-criteria decision-making method to systematically measure the quality of accessibility delivered to the client. This has resulted in a number of scientific publications.
Analysis of organizational e-Learning system (Achmea Company)
Mining techniques in cyber-attacks
Exchange rate prediction and inventory management
Supplier segmentation: Supplier segmentation is one of the strategic activities of sourcing and supply management of a firm. Rather than handling all suppliers individually, manageable number of segments are built. To capture the complicated interaction between different supplier aspects Mohsen was responsible for building a unified model to capture all available segmentation criteria. To ensure robustness of results, Mohsen proposed, implemented and tested an ensemble of classifiers that are constructed and combined and are then voted to give a final classification. The results encourage a group of colleagues to further extend the model and to test it on a number of case studies.
Mohsen also Contributed to courses at graduate and undergraduate level, including:
Methodologies: NN, SVM, PCA, Fuzzy logic, Evolutionary computation, Machine learning, single, multi and many-objective optimization, surrogate-assisted optimization, data-driven optimization, and decision making
TOUS STADT CO. CONSULTING ENGINEERS
SCIENTIFIC SOFTWARE DEVELOPMENT
|Jun. 2008 - Jan. 2009||Tous Stadt, among other activities, provides engineering consultancy specializing in design and supervisory services in field of transmission and sub transmission substation projects. As a software engineer Mohsen was responsible for implementation and testing of a Short Circuit Analysis software for Power Grid (substations) for internal use. He designed the tool and developed and tested it in Matlab. The tool was able to analyze the effect of 3-phase, 1-phase, line-to-line and line-to-ground fault currents on distribution substations.|
Islamic Azad University (IAU)
دانشگاه آزاد اسلامی
|Sep. 2006 - Jul. 2008||
Taught courses at undergraduate level, including:
Electrical Circuits I and II,
Electrical Measurement and Instrumentation,
Ferdowsi University of Mashhad -
Cognitive Computing Lab
|Sep. 2005 - Aug. 2008||Please have a look at my list of publications between 2005 and 2008.|
|2014||M. Davarynjead, Deploying Metaheuristics for Global Optimization, Delft University of Technology, Delft, The Netherlands, June 2014 (ISBN 978-90-5584-173-8)|
|2012||[pdf]||Mohsen Davarynejad, Jos Vrancken, Jan van den Berg, and Carlos A. Coello Coello, A Fitness Granulation Approach for Large-Scale Mechanical Optimization Problems, In Raymond Chiong and Zbigniew Michalewicz (Eds.), Variants of Evolutionary Algorithms for Real-World Applications, pp. 245-280, Springer, Berlin, 2012 (ISBN 978-3-642-23423-1).|
|[pdf]||Jan van den Berg, Guido van Heck, Mohsen Davarynejad, and Ron van Duin. Inventory Management, a Decision Support Framework to Improve Operational Performance, In Joao Eduardo Varajao and Maria Manuela Cruz-Cunha, editors, Organizational Integration of Enterprise Systems and Resources: Advancements and Applications. IGI Global, Hershey, PA 17033, USA, pp. 268-288, 2012.|
|2009||[pdf]||M. Davarynejad, S. Sedghi, M. Bahrepour, Chang Wook Ahn, M. Akbarzadeh, C. A. Coello Coello, Detecting Hidden Information from Watermarked Signal using Granulation Based Fitness Approximation, Applications of Soft Computing: From Theory to Praxis, Springer, Series: Advances in Intelligent and Soft Computing, Volume 58, pp. 463-472, 2009 (ISBN 978-3-540-89618-0).|
|2016||[pdf]||J. Rezaei, M. Davoodi, L. Tavasszy, M. Davarynejad, A multi-objective model for lot-sizing with supplier selection for an assembly system, International Journal of Logistics Research and Applications, pp. 125-142, 2016.|
|2014||[pdf]||M. Davarynjead, J. van den Berg, J. rezaei, Evaluating Center-Seeking and Initialization Bias: The case of Particle Swarm and Gravitational Search Algorithms, Information Sciences, Vol. 278, pp. 802-821, 2014|
|2012||[pdf]||J. van den Berg, M. Janssen, M. Davarynejad, and V. Marchau, Developing a System for Measuring Landside Accessibility Performance of Airports, Information and Communication Technologies for the Advanced Enterprise, pp. 65-85, 2012.|
|2009||[pdf]||M. Davarynejad, C.W. Ahn, J. Vrancken, J. van den Berg, C.A. Coello Coello, Evolutionary hidden information detection by granulation-based fitness approximation, Applied Soft Computing, Vol. 10(3), pp. 719-729, 2010, DOI: 10.1016/j.asoc.2009.09.001.|
|2008||[pdf]||M.-R. Akbarzadeh-T, M. Davarynejad, N. Pariz, Adaptive Fuzzy Fitness Granulation for Evolutionary Optimization, International Journal of Approximate Reasoning, Vol. 49(3), pp. 523-538, 2008, DOI Bookmark:10.1016/j.ijar.2008.05.004|
Selected Conference Papers:
|2015||[pdf]||Wolter Pieters and Mohsen Davarynejad, Calculating adversarial risk from attack trees: Control strength and probabilistic attackers, Data Privacy Management, Autonomous Spontaneous Security, and Security Assurance, Lecture Notes in Computer Science, Volume 8872, pp 201-215, 2015.|
|2012||[pdf]||Mohsen Davarynejad, Zary Forghany, Jan van den Berg, Mass-Dispersed Gravitational Search Algorithm for Gene Regulatory Network Model Parameter Identification, in 2012 Simulated Evolution And Learning (SEAL'12), Volume 7673 of Lecture Notes in Computer Science. pp. 62–72, 2012.|
|[pdf]||Zary Forghany, Mohsen Davarynejad, B. Ewa Snaar-Jagalska, Gene Regulatory Network Model Identification Using Artificial Bee Colony and Swarm Intelligence, in 2012 Congress on Evolutionary Computation (CEC'12), Brisbane, Australia, 10-15 June, 2012.|
|2011||[pdf]||Mohsen Davarynejad, Yubin Wang, Jos Vrancken, Jan van den Berg, Multi-phase Time Series Models for Motorway Flow Forecasting, the 14th International IEEE Conference on Intelligent Transportation Systems (ITSC'11), pp. 2033-2038, 2011.|
|[pdf]||M. Davarynejad, J. Rezaei, J. Vrancken, J. van den Berg and Carlos A. Coello Coello, Accelerating Convergence Towards the Optimal Pareto Front", in 2011 Congress on Evolutionary Computation (CEC'2011), New Orleans, pp. 2107-2114, 2011.|
|[pdf]||Maarten Janssen, Jan van den Berg, Mohsen Davarynejad, Vincent Marchau, Towards an Enterprise Information Subsystem for Measuring (Perceived) Landside Accessibility of Airports, International Conference on Enterprise Information Systems (CENTERIS 2011), Algarve, Portugal, pp. 72–81, 5-7 October 2011.|
|[pdf]||Mohsen Davarynejad, Jelmer van Ast, Jos Vrancken, and Jan van den Berg, Evolutionary Value Function ApproximationIEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL 2011) within the IEEE Symposium Series of Computational Intelligence (SSCI 2011), Paris, France, pp. 151-155, 11-15 April 2011.|
|[pdf]||Mohsen Davarynejad, Andreas Hegyi, Jos Vrancken, Jan van den Berg, Motorway Ramp-Metering Control with Queuing Consideration using Q-Learning, the 14th International IEEE Conference on Intelligent Transportation Systems (ITSC'11), pp. 1652-1658, 2011.|
|2010||[pdf]||Yubin Wang, Jos Vrancken, Mohsen Davarynejad, Implementing scenarios coordination for road network traffic control, IEEE International Conference on Systems, Man, and Cybernetics, Istanbul, Turkey, pp.230,234, 10-13 October 2010.|
|[pdf]||Yubin Wang, Jos Vrancken, Marcel Vale, Mohsen davarynejad, The scenario coordination module for the Municipality of Amsterdam and traffic management center of North-Holland, Proceedings of the 11th TRAIL Congress, 23 and 24 November 2010, (ISBN 978-90-5584-139-4).|
|[pdf]||Mohsen Davarynejad, Andreas Hegyi, Jos Vrancken, Yubin Wang, Freeway traffic control using Q-learning, Proceedings of the 11th TRAIL Congress, 23 and 24 November 2010, (ISBN 978-90-5584-139-4).|
|[pdf]||M. Davarynejad, S. Davarynejad, J. Vrancken, and J. van den Berg, Granular Value-Function Approximation for Road Network Traffic Control, International Conference on Networking, Sensing and Control, Chicago, USA, pp. 14-18, April 2010.|
|[pdf]||Guido van Heck, Jan van den Berg, Mohsen Davarynejad and Ron van Duin, Improving inventory management performance using a process-oriented measurement framework, International Conference on Enterprise Information Systems (CENTERIS 2010), Viano do Castelo, Portugal, pp. 279-288, 20–22 October 2010.|
|[pdf]||Yubin Wang, Jos Vrancken, Mohsen Davarynejad, Integration of Urban and Freeway Network Control by Using a Scenario Coordination Module, 13th International IEEE Conference on Intelligent Transportation Systems, Madeira Island, Portugal, pp. 671-676, 19-22 September 2010.|
|2009||[pdf]||Z. Forghany, M. Davarynejad, L. Zartash, Predicting the Impact of Supplemental Phytase, Wheat and Phosphorus on the Performance of Layin Hen, 7th World Congress on Computers in Agriculture, Reno, Nevada, pp. 233-238, 22-24 June 2009.|
|[pdf]||M. Davarynejad, J. Vrancken, A Survey of Fuzzy Set Theory in Intelligent Transportation: State of the art and future trends, Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics, San Antonio, Texas, pp. 4052-4058, Oct. 2009.|
|2008||[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T. and Carlos A. Coello Coello, Auto-Tuning Fuzzy Granulation for Evolutionary Optimization, in 2008 Congress on Evolutionary Computation (CEC'2008), pp. 3573-3580, IEEE Service Center, Hong Kong, June 2008. Slides for presentation as [[http://www.davarynejad.com/Resources1/WCCI2008.ppt|PowerPoint]] (3.4 MB).|
|[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T, N. Pariz, M. E. Golmakani, A.-R. khorsand, Fuzzy Surrogates in Fitness Function Approximation for Evolutionary Structure Design, 16th Iranian Conference of Electrical Engineering, Tehran, Iran, May 2008.|
|2007||[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T, N. Pariz, A-R khorsand, Adaptive Fuzzy Fitness Granulation in Structural Optimization Problems, Proceedings of the 2007 IEEE Multi-conference on Systems and Control (MSC'2007), pp. 172-177, Singapore, October 1- 3, 2007.|
|[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T, N. Pariz, A Novel General Framework for Evolutionary Optimization: Adaptive Fuzzy Fitness Granulation, Proceedings of the 2007 IEEE International Conference on Evolutionary Computing, (CEC'2007), pp. 951-956, Singapore, September 25-28, 2007.||[pdf]||M. E. Golmakani, K. Kamali, M.-R. Akbarzadeh-T., M. Kadkhodayan, M. Davarynejad, Application of a hybrid GA-BP optimized neural network for springback estimation in sheet metal forming process, in Proceedings of the First Joint Congress on Fuzzy and Intelligent Systems, Mashhad, Iran, August 29-31, 2007.||[pdf]||A. Jajarmi, R. Shahnaziand, A. Karimpour, M. Davarynejad, Designing a Robust controller for high-precision multi-dimensional positioning stage, (in Farsi) Tehran International Congress on Manufacturing Engineering (TICME), December 10-13, 2007.||2006||[pdf]||Ehsan Davarynejad, Mohsen Davarynejad, and GholamHossein Davarynejad, Year-Round Greenhouse Climate Control Using Evolutionary Fuzzy Systems, the 27th Int. Hortic. Cong. 13-19, Korea, August 2006.||2005||[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T., Time Series Prediction using Hierarchical Fuzzy Systems, In Proceedings of the 7th Conference of Intelligence Systems, pp. 202-207, Tehran, Iran, Dec. 2005.||2004||[pdf]||M. Davarynejad, M.-R. Akbarzadeh-T., A. Akramizadeh, Design, Simulation and Implementation of an Intelligent Greenhouse with Fuzzy Tools, In Proceedings of the Sixth Conference on Intelligent Systems, pp. 39-45, Kerman, Iran, Nov. 2004. (Nominated for Best paper award).|
List of Certifications
|2013||Discrete choice modeling||TRAIL research school, Delft University of Technology|
|2012||Mathematics of traffic||TRAIL research school, Delft University of Technology|
|2012||Microscopic traffic simulation model calibration and validation||TRAIL research school, Delft University of Technology|
|2011||TRAIL Theories & Methods||TRAIL research school, Delft University of Technology|