Soft Computing
- 1. Prof. Mohammad-R. Akbarzadeh-T Ferdowsi University of Mashhad A Presentation by: • Hosein Mohebbi • Bijan Nikkhah
- 2. Some Applications and Projects of Soft Computing Some Deep Learning Libraries Iranian Professors 2
- 3. 3
- 4. Google Google Translate Google Voice Search Gmail Inbox’s Smart Reply DeepDream IBM Microsoft LinkedIn Facebook Amazon Netflix Twitter SoundHound 4
- 5. 5
- 6. Kaggle Supervised by Geoffrey Hinton Emeritus Prof. Computer Science, U of Toronto & Engineering Fellow,Google 6
- 7. 7
- 8. IJCNN 8
- 9. varying level of difficulty : Spell Checking Keyword Search Finding Synonyms Parsing information from websites,documents, etc. Machine Translation Semantic Analysis Coreference Question Answering Stanford CoreNLP 9
- 10. 10
- 11. Text2Scene 11
- 12. BOT OR NOT Can you remember Turing Test? RKCP Algorithm: How it works? 12
- 13. Genetic Algorithm for Player Marking 13
- 15. Pros: • Lots of modular pieces that are easy to combine • Easy to write your own layer types and run on GPU • Lots of pre-trained models Cons: • You usually write your own training code (Less plug and play) • No commercial support • Spotty documentation (Facebook) Pros: • Python + Numpy • Computational graph abstraction, like Theano • Faster compile times than Theano • Data and model parallelism Cons: • Slower than other frameworks • Not many pre-trained models • Computational graph is pure Python,therefore slow • Drops out to Python to load each new training batch (Google) 15
- 16. Pros: • Good for feedforward networks and image processing • Good for fine-tuning existing networks • Train models without writing any code • Python interface is pretty useful Cons: • Need to write C++ / CUDA for new GPU layers • Not good for recurrent networks • Probably dying; slow development (Berkeley AI Research) Pros: • Python + Numpy • Computational graph is nice abstraction • High level wrappers (Keras, Lasagne) ease the pain Cons: • Raw Theano is somewhat low-level • Large models can have long compile times • Single GPU (Université de Montréal) 16
- 17. Pros: • Intuitive API inspired by Torch • Works with Theano and TensorFlow • Fast growing framework (François Chollet,A Google Engineer) Gensim is a fast implementation of word2vec implemented in Python. While Gensim is not a general purpose ML platform, for word2vec, it is at least an order of magnitude faster than TensorFlow. It is supported by the NLP consulting firm Rare Technologies. (RaRe Technologies) 17
- 19. Hossein Nezamabadi-pour h-index: 32 Professor of Electrical Engineering Shahid Bahonar University of Kerman • Swarm Intelligence • Image Processing • Pattern Recognition • Evolutionary Computation • Metaheuristics Mohammad-R.Akbarzadeh-T. h-index: 27 Professor of Electrical Engineering and Computer Engineering Ferdowsi University of Mashhad • Artificial Intelligence • Soft Computing • Fuzzy Logic • Complex Systems • Biomedical Engineering 19
- 20. Ali Aghagolzadeh h-index: 17 Professor of Electrical Engineering Babol Noushirvani University of Technology • Communication • Image processing • Computer Vision • WSN • Image Fusion Ataollah Ebrahimzadeh h-index: 19 Associate Professor Faculty of Electrical Engineering Babol Noushirvani University of Technology • Artificial Intelligence • WSN 20
- 21. Reza Ghaderi h-index: 20 Associate Professor Department of Electrical Engineering Shahid Beheshti University • Neural Networks • Artificial Inteligence • Pattern Recognition • Intelligent Control Mohammad Teshnehlab h-index: 27 Professor of Control Engineering K.N.Toosi University of Technology • Artificial Intelligence • Intelligent Control • Optimization • Interval Soft Computing 21
- 22. Azizollah Memariani h-index: 16 Professor of Mathematics Kharazmi University • Optimization • Multiple Criteria Decision Making • Fuzzy Systems • Decision Support Systems • Knowledge Engineering S. Mahmoud Taheri h-index: 20 Professor of Statistics University of Tehran • Fuzzy Systems • Statistical Inference 22
- 23. Behzad Moshiri h-index: 22 Professor of School of ECE University of Tehran Adjunct Professor of Univ. of Waterloo • Advanced Process Control • Sensor/Data Fusion • Mechatronics • Industrial Automation • Intelligent Transportation Systems Babak N Araabi h-index: 29 Professor of ECE University of Tehran • Machine Learning • Cognitive Modeling • MachineVision • System Identification 23