Biomedical and Health Informatics Team (BioHIT)

Printer-friendly versionSend by email

About BioHIT

Biomedical and Health Informatics Team (BioHIT) is a meeting team where members and invited speakers give presentations and participate in conversations about biomedical and health informatics, bioinformatics and related fields.

BioHIT meetings begun on April 2016 (initially as BioCAKT meetings) and since then a new presentation take place in Institute of Informatics and Telecommunications, NCSR D, twice a month.

Previous presentations details and relative material are available in this page.

BioHIT seminar presentations

  • Data augmentation as a biologically plausible alternative to explicit regularization in CNNs

    The impressive success of modern deep neural networks on computer vision tasks has been achieved through models of very large capacity compared to the number of available training examples. This over-parameterization is often said to be controlled by means of different regularization techniques, mainly weight decay and dropout. However, since these techniques reduce the effective capacity of the model, typically even deeper and wider architectures are required to compensate for the reduced capacity. Therefore, there seems to be a waste of capacity in this practice. In contrast, data augmentation techniques do not reduce the effective capacity and improve generalization by increasing the number of training examples. This talk will present the results of an ablation study on some popular architectures that conclude that data augmentation alone—without any other explicit regularization techniques—can achieve the same performance or higher than regularized models, especially when training with fewer examples, and exhibits much higher adaptability to changes in the architecture. Besides, a recent study suggests that models trained with heavier data augmentation exhibit more similarity with the human inferior temporal (IT) cortex.

    Alex Hernández-García, 8/10/2018 (Slides)

    Alex Hernández-García is a third-year PhD candidate at the Neurobiopsychology research group at the Institute of Cognitive Science of the University of Osnabrück, Germany. He obtained his B.Sc. in Audiovisual Systems Engineering in 2014, as well as his M.Sc. in Computer Vision and Machine Learning in 2015 from the University Carlos III of Madrid, Spain. At this university, he was a research assistant at the Signal Theory and Communications Department from 2013 until February 2016, when he moved to Berlin, Germany, to pursue his PhD with a Marie Sklodowska-Curie ITN fellowship. His research interests range from computer vision and machine learning to visual perception and computational neuroscience. In particular, he is currently interested in exploring the connections between artificial neural networks and the visual processing in the primate brain.

  • Spontaneous transitions of functional connectivity and modular structure in endogenous brain activity (click for more)

  • Machine learning opportunities for a new generation of microwave-based medical devices (click for more)

  • Structured Element Search for Scientific Literature (click for more)

  • Αόρατες Πόλεις. Δομή της χρωματίνης και γονιδιωματική αρχιτεκτονική σε ευκαρυωτικά γονιδιώματα (click for more)

  • HEALTH BANK – A Workbench for Data Science Applications in Healthcare (click for more)

  • Μελέτη αναπαραστάσεων γονιδιωματικών ακολουθιών σε προβλήματα ταξινόμησης (click for more)

  • The In Silico Oncology and In Silico Medicine Group of ICCS-SECE-NTUA and the Large Scale EU-US Research Project CHIC on In Silico Oncology (click for more)

  • Enabling reproducibility in critical care research (click for more)

  • Atom Mapping of Chemical Reactions (click for more)

  • Γενετική του καρκίνου μαστού-ωοθηκών (click for more)

  • RADIO: Unobtrusive, Efficient, Reliable and Modular Solutions for Independent Ageing (click for more)

  • The Thalamic Visual Prosthesis Project (click for more)

  • Όταν το p-value εξορίστηκε ή τι να ΜΗΝ κάνω όταν κάνω έρευνα (click for more)

  • CRISPR/Cas based technology : Taking Gene Editing to the next level (click for more)

  • Drug Repositioning: Existing approaches and applicability for Duchenne Muscular Dystrophy (click for more)

  • Approaches to the atom mapping problem: A review (click for more)

  • Ευρετηρίαση ακολουθιών DNA με τη χρήση χαρακτηριστικών από γράφους Ν-γραμμάτων (click for more)

  • Using Text Mining to Discover Repositioning Opportunities for Orphan Drugs and Rare Diseases (click for more)

  • Machine learning classification of autopsy proven vascular cognitive impairment and Alzheimer's disease using connected speech sampling (click for more)

  • BioASQ: "A challenge on large-scale biomedical semantic indexing and question answering" (click for more)

  • Reaction Map: An efficient atom mapping algorithm for chemical reactions (click for more)

© 2019 - Institute of Informatics and Telecommunications | National Centre for Scientific Research "Demokritos"

Terms of Service and Privacy Policy