About

I am a Neuroscience Ph.D. candidate at George Mason University (GMU) working with Dr. Giorgio Ascoli at the Center for Neural Informatics, Structures, and Plasticity (CN3). Currently I work as a graduate research assistant at Krasnow Institute for Advanced Studies. My research interests include various aspects of Computational Neuroscience, Neuroinformatic, Deep Learning, and Natural Language Processing.

At the moment, my research focuses on the administration, facilitation, and effortless extraction of neuroscience metadata by leveraging web-technologies and machine learning approaches. Moreover, by employing computer science and neuroscience mentality, we are trying to enhance the quality in results, robustness in responses, and generalization in the models that spring from artificial neural networks based on biological connectivity of the real brain.

Prior to the GMU, I had the opportunity to be a master student at the University of Tehran, Iran. Please contact me via kbijari[at]gmu[dot]edu.

What's new?

NEW! Journal paper: An open-source framework for neuroscience metadata management applied to digital reconstructions of neuronal morphology [Link], [GitHub]

NEW! Journal paper: Leveraging deep graph-based text representation for sentiment polarity applications [Link]

Research & Academic services

Research

My research interests include multiple aspects of neuroinformatics, natural language processing, and deep learning. Main contributions and current research focus on:

NeuroMorpho.Org is an online repository of over thousands of digitally reconstructed neurons and glia shared by laboratories worldwide. Every entry of this public resource is associated with essential metadata describing animal species, anatomical region, cell type, experimental condition, and additional information relevant to contextualize the morphological content. Administration, facilitation, and management of metadata entries for these reconstructions is a project I am undertaking.

We have recently transitioned NeuroMorpho.Org has from its original manual spreadsheet-based metadata annotation system to a custom-developed, robust, web-based framework for extracting, structuring and managing neuroscience information. This conversion facilitated metadata annotation, improved terminology management, and made NeuroMorpho.Org one step closer to its aim that is to minimize the barrier for direct knowledge sharing by domain experts.

An important goal in neuroinformatics is the systematic extraction of technical information from neuroscience publications. The traditional task of manual annotation can be difficult, error-prone, time-consuming, and labor-intensive. Thus, progressively automating the key elements of this process is highly desirable. Here, we aim to introduce a novel named entity recognition system to extract neuroscience name entities from published articles based on state-of-the-art deep neural networks. Our method leverages unsupervised and supervised learning schemes and is independent of hand-crafted features, primarily leveraging contextual information. The specific application of this development is to facilitate metadata administration and curation for NeuroMorpho.Org.

Perhaps one of the most intriguing intersections in science is where neuroscience and computer science meet. Knowing how brains function and the mystery behind consciousness have always been questions for humans. As neuroscience advances in this endeavor cell by cell and neuron by neuron, artificial intelligence is getting better at performing human tasks such as prediction, recognition, and repetition. However, most of AI models, particularly the massive neural networks, are extremely task-specific, poorly generalizable, and time-consuming to train. To bring such large models back on the actual track and a step closer to the biological reality, we inspire and study physiological facts and order them in artificial neural networks to make robust, generalizable, and efficient models.

Journal reviewer

Neuroinformatic
Expert Systems with Applications

Teaching

(TA) Cellular, Neurophysiological, and Pharmacological Neuroscience
(TA) Pattern Recognition and Machine Learning
(TA) Statistical Machine Learning
(TA) Theory of Language and Automata
(TA) Fundamentals of Computer Programming

Selected publications

K Bijari, MA Akram, GA Ascoli "An open-source framework for neuroscience metadata management applied to digital reconstructions of neuronal morphology", Brain Informatics 7 (1), (2020)
K Bijari, H. Zare, E Kebriaei, H Veisi "Leveraging deep graph-based text representation for sentiment polarity applications", Expert Systems with Applications (144), (2020)
K Bijari, H. Zare, H Veisi, H Bobarshad "Memory-enriched big bang–big crunch optimization algorithm for data clustering", Neural Computing and Applications (29), (2018)
E Gharavi, K Bijari, K. Zahirnia, H Veisi "A deep learning approach to persian plagiarism detection", Fire working notes, (2016)

News

New Journal Paper 26/Mar/2020
We have implemented an open-source metadata administration framework as an extension to NeuroMorpho.Org to assist metadata management for reconstructions of neuronal morphology. This work is published in BrainInfromatics.
New Journal Paper 15/Nov/2019
A new paper titled "Leveraging deep graph-based text representation for sentiment polarity applications" that focuses on text analysis from a graph-based perspective is published in Expert Systems with Applications.
Poster Presentation 11/Apr/2019
I have presented "Computer-Assisted Metadata Annotation of Digitally Reconstructed Neuronal Morphology" at the 5th Annual BRAIN Investigators Meeting.