Hamid Palangi Electrical and Computer Engineering Department The University of British Columbia Office: ICICS/CS X310, 2366 Main Mall Vancouver, BC V6T 1Z4, Canada Research Interests Email: hamidp@ece.ubc.ca http://ece.ubc.ca/∼hamidp Machine Learning, Deep Learning Linear Inverse Problems Applications in Text and Image data Research As a Research Intern at Microsoft Research, Redmond, WA (May 2014 - August 2014) and Work · Worked on Ranking for Web Search using Context information in queries and documents which Experience resulted in Deep Structured Semantic Modeling with Recurrent Neural Networks (RDSSM) and Long Short Term Memory (LSTM-DSSM). (Derivation + Implementation in C# and CUDA). This work resulted in the following paper: – Hamid Palangi, Li Deng, Yelong Shen, Jianfeng Gao, Xiaodong He, Jianshu Chen, Xinying Song, Rabab Ward, “Deep Sentence Embedding Using the Long Short-Term Memory Networks: Analysis and Application to Information Retrieval ”, to appear in IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2016 As a Research Intern at Microsoft Research, Redmond, WA (June 2013 - August 2013) · Worked on designing and implementing Recurrent Deep Stacking Networks for Automatic Speech Recognition. (Derivation + Implementation in MATLAB). This work resulted in the following papers: – Hamid Palangi, Li Deng, Rabab Ward “Learning Input and Recurrent Weight Matrices in Echo State Networks”, in NIPS Workshop on Deep Learning, Lake Tahoe, Nevada, USA, December 2013 – Hamid Palangi, Li Deng, Rabab Ward, “Recurrent Deep-Stacking Networks for sequence classification”, in IEEE China Summit & International Conference on Signal and Information Processing (ChinaSIP), Xi’an, 2014. As a Research Assistant at Image and Signal Processing Lab, ECE Department, The University of British Columbia, Vancouver, Canada (Jan. 2012 - now) · Working on Deep Learning Methods for Sequence Modelling: Applications to Compressive Sensing and Sentence Embedding As a Research Assistant at Digital Signal Processing (DSP) Lab, Sharif University of Technology, Tehran, Iran (Sept. 2007 - Jan. 2010) · Worked on Sparse Decomposition and Mixed Transform Techniques for Signal Compression Publications Journal Papers: 1. Hamid Palangi, Li Deng, Yelong Shen, Jianfeng Gao, Xiaodong He, Jianshu Chen, Xinying Song, Rabab Ward, “Deep Sentence Embedding Using the Long Short-Term Memory Networks: Analysis and Application to Information Retrieval ”, to appear in IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2016 2. Hamid Palangi, Rabab Ward, Li Deng, “Distributed Compressive Sensing: A Deep Learning Approach”, under review in IEEE Transactions on Signal Processing, 2016 3. Mohammad S. E. Abadi, Hamid Palangi, “Mean-Square Performance Analysis of the Family of Selective Partial Update and Selective Regressor Affine Projection Algorithms”, Elsevier Signal Processing, Volume 90, Issue 1, January 2010, Pages 197-206 4. Hamid Palangi, Mohammad H. Refan, “Error Reduction of a Low Cost GPS Receiver for Kinematic Applications Based on a New Kalman Filtering Algorithm”, International Journal of Innovative Computing, Information and Control (IJICIC), Vol.6, No.8, August 2010, Pages 3775-3786 Hamid Palangi II 5. Mohammad S. E. Abadi, Hamid Palangi, “A Unified Approach to Set-Membership and Selective Partial Update Adaptive Filtering Algorithms”, International Journal of Information and Communication Technology (a journal published by Iran Telecommunication Research Center (ITRC)), Vol. 2, No. 2, July 2010, Pages 61-70 Conference Papers: 6. Hamid Palangi, Rabab Ward, Li Deng, “Exploiting Correlations Among Channels in Distributed Compressive Sensing with Convolutional Deep Stacking Networks”, to appear in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), China, 2016 7. Hamid Palangi, Li Deng, Rabab Ward, “Recurrent Deep-Stacking Networks for sequence classification”, in IEEE China Summit & International Conference on Signal and Information Processing (ChinaSIP), Xi’an, 2014. 8. Hamid Palangi, Li Deng, Rabab Ward “Learning Input and Recurrent Weight Matrices in Echo State Networks”, in Deep Learning Workshop in Conference on Neural Information Processing Systems (NIPS), Lake Tahoe, Nevada, USA, December 2013 9. Hamid Palangi, Rabab Ward, Li Deng, “Using Deep Stacking Network to Improve Structured Compressed Sensing with Multiple Measurement Vectors”, in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Vancouver, 2013. 10. Hamid Palangi, Aboozar Ghafari, Masoud Babaie-Zadeh, Christian Jutten, “Image Coding and Compression with Sparse 3D Discrete Cosine Transform”, Lecture Notes In Computer Science; Vol. 5441, Proceedings of the 8th International Conference on Independent Component Analysis and Signal Separation, Paraty, Brazil, 2009, Pages 532-539 11. Hamid Palangi, Shohreh Kasaei, “Fast and Robust Multi-Frame Super-Resolution Using Inhibition Principle”, ICDT, Fourth International Conference on Digital Telecommunications, France, 2009, Pages 82-87 12. Aboozar Ghafari, Hamid Palangi, Masoud Babaie-Zadeh, Christian Jutten, “ECG Denoising and Compression by Sparse 2D Separable Transform with Over Complete Mixed Dictionaries”, IEEE International Workshop on Machine Learning for Signal Processing, Grenoble, France, 2009, Pages 1-6 Education Ph.D., Electrical and Computer Engineering, The University of British Columbia,Vancouver, Canada. (January 2012 - present), GPA: 93.5 / 100 via 6 credits · Co-advised by Prof. Rabab Ward and Prof. Li Deng · Graduate Courses: Imaging and Estimation with Wavelets (94 / 100, first mark), Multimedia Systems (93 / 100, first mark), Linear Inverse Theory (audit) · Thesis: Deep Learning Methods for Sequence Modelling: Applications to Compressive Sensing and Sentence Embedding Deep Learning Summer School, Universit de Montreal, Montreal, Canada. (Aug.03 to Aug.12, 2015), Organizers: Yoshua Bengio, Roland Memisevic, Yann LeCun M.Sc., Electrical Engineering, Sharif University of Technology, Tehran, Iran. (September 2007 - January 2010), GPA: 18.01 / 20 via 43 credits · Advisor: Dr. Massoud Babaie-Zadeh · Graduate Courses: Random Processes, Digital Signal Processing, Adaptive Filters, Data Networks, Advanced Image Processing (20 / 20, first mark), Digital Video Processing (20 / 20, first mark), Special Topics in Communications (BSS and Sparse Signal Processing), Neural Networks (20 / 20, first mark), Electronic Interface Circuits, Advanced Object Oriented Programming, Advanced VLSI Design, Advanced Microprocessors, DSP Processors (20 / 20, first mark) · Thesis: Sparse Decomposition and Mixed Transform Techniques for Signal Compression B.Sc., Electrical Engineering, Shahid Rajaee University, Tehran, Iran, (September 2003 - September 2007), GPA: 18.54 / 20 via 151 credits · Advisor: Dr. Mohammad Hossein Refan · Thesis: Error reduction of a low cost GPS receiver using a new Kalman filtering algorithm for kinematic applications Hamid Palangi III Honors and Selected as one of 7 students who were exempted from PhD entrance exam of Sharif University Awards of Technology for Electrical Engineering, February 2011 Ranked 3rd in my M.Sc. class Ranked 1st in my B.Sc. class Ranked 4th in national automation M.Sc. entrance exam among about 2000 people and offered a scholarship by petroleum ministry, September 2007, not attended Ranked 33rd in national Electrical Engineering M.Sc. entrance exam among about 17000 people, September 2007 Teaching The University of British Columbia as a Teaching Assistant in: Experience · EECE 251: Circuit Analysis I (Undergraduate Course, Fall 2012), EECE 259: Introduction to Microcomputers (Undergraduate Course, Fall 2012), EECE 253: Circuit Analysis II (Undergraduate Course, Winter 2013), EECE 261: Engineering Electromagnetics (Undergraduate Course, Winter 2013) Sharif University of Technology as a Teaching Assistant in: · Computer Architecture (Undergraduate Course, Fall 2011), Computer Architecture Lab (Undergraduate Course, Fall 2011), Advanced Digital Signal Processing(Graduate Course,Spring 2009), Digital Video Processing(Graduate Course,Fall 2009), Pulse and Digital Circuits (Undergraduate Course,Spring 2009), Introduction to Electrical Engineering(Undergraduate Course,Spring 2009) Shahid Rajaee University as a Teaching Assistant in: · Electromagnetics(Undergraduate Course,Fall 2005 and Fall 2006), Electronic Circuits I(Undergraduate Course,Fall 2005), Electronic Circuits II(Undergraduate Course,Spring 2005), Linear Control Systems(Undergraduate Course,Spring 2007) Projects and Reports Exploiting Probabilistic Joint Sparsity in Distributed Compressive Sensing: I got the first mark for this project in EOSC513 course at UBC (Design and implementation in MATLAB) Coarse Grain Quality Scalable High Efficiency Video Coding (HEVC): Group project, we got the first mark for this project in EECE541 course at UBC (Design and implementation in C++, we worked thoroughly on HM5 source code in this project) Robust and Fast Multi-frame Image Super Resolution with Inhibition Principle (Design and implementation in MATLAB) Clustering of up to 100 sources of information without any primary knowledge of the number of sources or features for huge amount of data (Design and implementation in C++) Mixed transform techniques for image compression (Design and implementation in MATLAB) Cichocki Neural networks to solve ill-posed linear system of equations (Design and implementation in MATLAB SIMULINK) A thorough experimental study about error of various generations of GPS satellites and developing a thorough MATLAB toolbox for an improved OEM electronic board data analysis (as part of my B.Sc. project) Implementation of a real time sound stretching algorithm on Texas Instrument TMS320C6416 DSP processor (as my DSP Processors course project) Presentations On Deep Structured Semantic Modeling with Recurrent Neural Networks (RDSSM) and Long Short Term Memory (LSTM-DSSM), presented in Microsoft Research (Redmond), August 2014 On Recurrent Deep Stacking Networks for Speech Recognition, presented in Microsoft Research (Redmond), August 2013 On Structured Compressive Sensing, presented in the course EOSC513, March 2012 On Coarse Grain Quality Scalable High Efficiency Video Coding (HEVC), presented in the course EECE541, April 2012 On Sparse Decomposition and Mixed Transform Techniques for Signal Compression (M.Sc. Thesis Defense), January 2010 A new Kalman filtering algorithm for GPS positioning for kinematic applications (B.Sc. Thesis Presentation), September 2007 Hamid Palangi IV A new approach based on inhibition principle and Cichocki neural network for over determined scenario of Super Resolution problem (as my Advanced Image Processing course project), March 2009 Neural Networks for Solving Linear System of Equations and their Analog Implementation (Neural Networks course project), December 2008 Skills Programming: C#, CUDA, MATLAB, Familiar with Data Structures and Algorithms, Familiar with C++ and Python Standards: HEVC/H.265 Encoder and Decoder HM (in C++)