Ramtin Mehdizadeh Seraj

School of Computing Science
Simon Fraser University

I am a Master student of Computing Science at Simon Fraser University and affiliated with Natural Language Laboratory (Natlang), where I am very fortunate to work under the supervision of Dr. Anoop Sarkar and Dr. Fred Popowich.

Simply put, I am passionate about accuracy and scalability of Machine Learning approaches especially in applications related to Natural Languages.
One of these applications are Statistical Machine Translation which I am working on right now. Many challenges exist in this application to achieve a high quality (accurate) translations; the first challenge that I'm trying to tackle, using semi-supervised methods, is finding translation for unseen words/phrases. Another challenge is to leverage multilingual resources (e.g. bilingual embeddings) for Machine Translation.

Moreover, many Machine Learning approaches (like Deep Learning methods) are not suitable for huge data sets (Big data). My second area of interest, is scaling up these methods, which can be done using distributed computing frameworks (e.g. Hadoop, Spark) or exploiting special computing units (e.g. GPU's).

We are drowning in information but starving for knowledge. - John Naisbitt


Master of Science (Fall 2013 - present)

School of Computing Science
Simon Fraser University
8888 University Drive
Burnaby, BC V5A 1S6, Canada

Bachelor of Engineering (Fall 2008 - Summer 2013)

Department of Computing Engineering
Amirkabir University of Technology
Thesis : Unsupervised Named-Entity Recognizer for Farsi
Supervisor : Dr. Shahram Khadivi

All models are wrong, but some are useful. - George Box


Improving Statistical Machine Translation with a Multilingual Paraphrase Database
Ramtin Mehdizadeh Seraj, Maryam Siahbani and Anoop Sarkar. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP 2015). Sep 17-23, 2015. Lisbon, Portugal.

Incremental Translation using a Hierarchical Phrase-based Translation System.
Maryam Siahbani, Ramtin Mehdizadeh Seraj, Baskaran Sankaran and Anoop Sarkar. In Proceedings of the 2014 IEEE Spoken Language Technology Workshop (SLT 2014). December 7-10, 2014. Nevada, USA.

A Novel Unsupervised Method for Named-Entity Identification in Resource-poor Languages Using Bilingual Corpus
Ramtin Mehdizadeh Seraj, Fattaneh Jabbari and Shahram Khadivi. In Proceedings of the 2014 International Symposium on Telecommunications (IST 2014). September 9-11, 2014. Tehran, Iran.

An Analysis of Affective Words in Machine Translation
María Aloy Mayo, Ramtin Mehdizadeh Seraj, Ana P García Varela, Dan Fass, Fred Popowich, Anoop Sarkar and Maite Taboada. In Proceedings of the 3rd Pacific Northwest Regional NLP Workshop: (NW-NLP 2014). April 2014. Redmond, WA, USA.


3rd Pacific Northwest Regional NLP Workshop - Microsoft Research

Hope for the future is our most expressive protest. - Mirhosein Mousavi

Code Repository

Graph Propagation for MT

My research repository.

Large Scale NLP

Recently, I have tried to use Apache Spark for some NLP tasks.

Deep Codes

To satisfy my curiosity, I am using Deep Learning for NLP. Most of my codes are based on Theano.

Natlang Matrix

Where we store our baseline MT scores in our lab.

Bibtex Viewer

Bibtex viewer javascript code (based on bibtex-js )

Be the change that you wish to see in the world. - Mahatma Gandhi


My Machine Learning Knowledge Graph

I have tried several times to come up with a categorization of materials in Machine Learning that I am familiar with, and still it looks similar to " Celestial Emporium of Benevolent Knowledge " categorization of animals. So it is still an open challenge for me, any suggestion?

SFU - Machine Learning Reading Group

A reading group that meets weekly to discuss current advanced topics in Machine Learning.

Twenty Years of Bitext

An annotated transcript of Bob Mercer and Peter Brown's talk at the EMNLP workshop 2013 by Matt Post.

Computer-related Milestones in My Life

  • 1990 - Hello world!
  • 1996 - Playing "Loderaner" on my mother's Commodore 64 (Big Achievement)
  • 2000 - I understood there exists another type of telephone for computers: Dial up Internet
  • Sep 2006 - I found that Statistics might help in real life, but I was not sure
  • August 2008 - I realized that I am not a lucky person, So I decided to become an Computer Engineer to fix that
  • Sep 2008 - I wrote my first program in C "Paint with Draw and Fill"
  • Feb 2009 - For the first time I considered myself as an object with attributes and methods
  • June 2010 - Facing my first problem with the size of data (Processing 40 GB of Text, My solution: waiting for a long time.)
  • July 2010 - My labtop died
  • Mar 2011 - We are not alone, Machines can also learn
  • April 2011 - Traning my first statistical model (linear regression, Ta-Da)
  • June 2012 - Joining NLP Lab, Amirkabir University working with Dr. Khadivi
  • Feb 2012 - Sending my first paper to a conference
  • March 2012 - My first paper got rejected !
  • Sep 2013 - Moving to Canada, Having coffee with Anoop
  • Jan 2014 - Facing my second problem with the size of data (I was not patient enought to do my previous solution, so I decided to learn about Hadoop)

Some of my long term goals

Going to Antarctica
Having impact on the educational systems to reduce ignorance
Doing Heli-Skiing in Whistler
Climbing Everest
Solving a machine learning problem on a quantum computer like D-Wave two

In loving memory of my grandfather, my hero