Mr Jiri Fajtl

Research project: Visual Memories: Memorization and Recall of Visual Information in Artificial Neural Networks


Within this work we set to advance research in the domain of neural memory systems, focusing particularly on the issues of catastrophic interference, few-shots and continual learning, autoassociative and predictive episodic memorization and retrieval. The progress in these areas is imperative to the realization of true intelligent agents. While the ability to acquire, retain and recall past experiences comes natural to us, it represents a long standing challenge for machine learning systems, namely being able to continually learn from new information coming from non stationary environments without interfering with the past knowledge, associating new observations with the old ones, predicting temporal sequences, transferring old knowledge to new situations or generating fictional memories derived from the accumulated world knowledge.

  • Research degree: PhD
  • Title of project: Visual Memories: Memorization and Recall of Visual Information in Artificial Neural Networks
  • Other research supervisor: Professor Vasilis Argyriou


With an engineering background in electronic and radio communication, I spent more than 20 years in the industry conducting research and development in domains ranging from DSP, peer to peer networking, dynamic systems and control, computer vision and machine learning, primarily for robotics. Over the time I have started several companies where I leveraged my passion for the multidisciplinary  software and hardware engineering. Very recently I have been offering my expertise as an independent contractor/consultant. I hold MSc from Kingston University, London in embedded systems and computer vision, and currently pursuing PhD in machine learning for computer vision.

Areas of research interest

  • Artificial Inteligence
  • Machine Learning
  • Computer Vision


  • MSc Embedded Systems (Computer Vision), Kingston University, London


  • Dupre R, Fajtl J, Argyriou V, Remagnino P. Improving dataset volumes and model accuracy with semi-supervised iterative self-learning. IEEE Transactions on Image Processing. 2019 May 6; 29:4337-48.

Conference papers

  • Fajtl J, Argyriou V, Monekosso D, Remagnino P. Amnet: Memorability Estimation with Attention. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2018) (pp. 6363-6372).
  • Fajtl J, Sokeh HS, Argyriou V, Monekosso D, Remagnino P. Summarizing Videos with Attention. In Asian Conference on Computer Vision (ACCV 2018)  (pp. 39-54). Springer
  • Fajtl J, Argyriou V, Monekosso V, Remagnino P. Latent Bernoulli Autoencoder.  In International Conference on Machine Learning (ICML 2020) 37, 5524-5534
  • Mandal B, Fajtl J, Argyriou V, Monekosso D, Remagnino P. Deep residual network with subclass discriminant analysis for crowd behavior recognition. In2018 25th IEEE International Conference on Image Processing (ICIP 2018) (pp. 938-942). IEEE.