Seminars 2021

Date: December 24th, 2021
Speaker: Latypov Insan-Aleksandr, HSE, Moscow
Topic:"Neuro-Symbolic approaches in Visual Question Answering problems" 
Abstract: In this presentation, we continue to demonstrate possible solutions in solving Visual Question Answering tasks using Neuro-Symbolic approaches. The discussed MAC architecture utilizes the attention mechanism to focus on the right answer. On the other hand, Neuro-Symbolic Concept Learner uses Domain Specific Language to answer the questions. In this seminar we will discuss the Neural State Machine which is based on traversing the knowledge graphs and operating with the finite set of concepts to perform reasoning.

Date: December 17th, 2021
Speaker: Latypov Insan-Aleksandr, HSE, Moscow
Topic:"Neuro-Symbolic approaches in Visual Question Answering problems" 
Abstract: In this presentation we demonstrate possible solutions in solving Visual Question Answering tasks using Neuro-Symbolic approaches. This paradigm aims to bridge the gap between symbolic and connectionist models. Visual Question Answering implies answering the questions based on the given picture and requires reasoning over the extracted information. Models, that we review during this talk, solve the problem of reasoning in different ways. Some of them are based on traversing the knowledge graphs, while others translate the concepts to executable, symbolic programs.

Date: November 26th, 2021
Speaker: Attila Kertesz-Farkas, HSE, Moscow
Topic: "Computational methods for tandem mass spectrometry data annotation"
Abstract: In this talk we present rencently published computational methods and statistical protocols in order to improve the number for spectrum annotation annotated with high confidence. The methods presented can be categorized in few main areas as follows. First, we present new spectrum scoring methods with improved discriminative power. Then we present statistical methods for score calibration or statistical protocols with increased statistical power in spectrum annotation. Then we present methods which find post-translational modifications in the spectrum data. Finally, we present spectrum filtering methods.

Date: October 1st, 2021
Speaker:  Polina Kudryavtseva, HSE, Moscow
Topic: "Deep Convolutional Neural Networks Help Scoring Tandem Mass Spectrometry Data in Database-Searching Approaches
Abstract: The talk is about deep convolutional neural network that increases the number of spectrum annotations in database-searching-based systems. The developed model is stable and fast to train and it slightly outperforms the current state-of-the-art methods. Participants discussed the architecture of the model, methods for evaluating it, and the results achieved on low and high resolution datasets. This work was published in the Journal of Proteome Research in 2021

Date: June 29th, 2021
Speaker: Attila Kertesz-Farkas, HSE, Moscow
Topic: "Introduction to computational mass spectrometry"
Abstract: The general issues of the application of tandem mass spectrometry, spectrometers and data interpretation were outlined, as well as detailed information about computational methods and protocols for analyzing and verifying annotations of tandem mass spectrometry data

Date: June 25th, 2021
Speaker: Attila Kertesz-Farkas, HSE, Moscow
Topic: "Introduction to computational mass spectrometry"
Abstract: The general issues of the application of tandem mass spectrometry, spectrometers and data interpretation were outlined, as well as detailed information about computational methods and protocols for analyzing and verifying annotations of tandem mass spectrometry data