Fatemeh Sadat Rezvaninejad

I am a master student at Iran University of Science and Technology, Department of Computer Engineering (IUST), which is one of the best universities in Iran. I work under Prof. Mohammad Reza Kangavari's supervision.

I have received my bachelors in Computer Engineering from Shahid Bahonar University of Kerman(SBUK)

I am interested in Natural Language Processing, Recommendation Systems, Sentiment Analysis, Deep Learning and Transfer Learning.

Email  /  CV  /  LinkedIn  /  Google Scholar

Some of my researches and academic projects are shown below.

Persian Stance Classification Dataset
TTO 2019
In this paper we present the first stance detection dataset in Persian which has applications in fact-checking and summarization (Ferreira and Vlachos, 2016). We developed a web-based tool for importing rumored claims, collecting associated news-articles and labeling their stance against the claims. In addition we introduce language specific features that outperform all baseline systems on this dataset.

Analyzing Efficiency of GPU for Executing High Performance Computing Applications
ICSETI 2017
The need for computer engineers to do graphic work of high computing time such as photoshop, animation and etc. has also been instrumental in understanding and developing the CUDA parallel graphics and graphics processor.In this paper we try to explain the role and importance of Cuda programming in the knapsack problem and image processing.

Sentence-State lstm for Text Presentation
Data Mining 2020

In this project I worked on "Sentence-State lstm for Text Presentation" paper which has shown an alternative LSTM structure for encoding text, which consists of a parallel state for each word. This paper has shown that the proposed model has strong representation power, we used some augmentation techniques to augment data which helps to reduce overfitting and ... .

Interactive Machine Learning( Human in the Loop)
Seminar 2019

In this presentation and seminar I talked about Interactive Machine Learning. Interactive Machine Learning are defined as algorithms which iteracts with agents which can be humans and the interactions will be optimised. The benefits of using Interactive Machine Learning is that it doesn't need an expert person in the learning process and user can participate in the learning process in the easies way.

Language Model on Eda(Easy Data Augmentation)
Deep Learning 2019

Having a small dataset for training is a necessary problem in many fields of deep learning, so we decided to investigate how we can augment it. We investigated on ways that introduced before on this problem and find other ways of augmenting data such as oppositing some words of text and for not changing the labels we added not, an other way that we suggest is to randomly deleting some words and randomly replacing some words in the text with their synonyms. For the end of the work we gave these into a Language Model to find out if it improves the predictions or not.

we have reviewed "A Simple and Effective Approach to the Story Cloze Test paper" in which the authors have introduced their approach to the Story Cloze Test. We follow (Srinivasan et al., 2018) approach to achieve the same result and discuss how we can improve it. we used pre-trained skip-thought encoder1 and glove 2. our main code consists of preprocessing and skipthought-cloze-valsoftmax to achieve the same result as (Srinivasan et al., 2018) achieved.

That was cool!