Deep Learning in Computer Vision

We research novel neural network architectures and approaches to various computer vision tasks. The tasks include human-focused action detection in videos, image classification and optical flow estimation. For more information see the Showmaxlab website.

Recurrent neural networks

We improve artificial neural networks for predictive modeling but also for reinforcement learning or sequence modeling. For the later applications, recurrent neural networks are better fit. We work on state of the art architectures.

Big Data, Scalable and Robust Computational Infrastructures

Extending Apache Mesos, Apache Spark, Storm, Kafka, Cassandra, Zookeeper, ... for big data processing, extensive computation, robust deployment and maintenance. Implementation of scalable AI algorithms for big data platforms (Apache Spark, Hadoop).

Relational Machine Learning

Our goal is to automate transformation of relational data (think about SQL database) into a single table of features, which can be used for classification, regression, clustering, outlier detection... And by this automation alleviate the biggest hurdle in the process of data mining - data preprocessing.

Read more: Relational Machine Learning

Advanced Data Clustering

Optimization and multi-objective evaluation of clustering algorithms. Time series clustering and interactive evolution of clustering.

Decision Making under Uncertainty

Research and development of intelligent agent for solving problems in a complex (continuous) environments. Agent has to make decisions with incomplete information, moreover under uncertainty about consequences of his moves. (Agent used in autonomous playing of Angry Birds.)

Multi-agent Simulation of Behaviour

Multi-agent systems are traditionally used for simulation of systems with a huge amount of interacting entities. Its dynamic is usually too complex but the behaviour of entities can be individually described. We analyse and extract behavioural patterns from available data and simulate complex dynamic systems in multi-agent environment.

Recommender systems

Advanced general recommendation algorithms for items such as products, movies, songs, articles, etc. Research into ensembles of recommender systems, metalearning in recommendation, multiobjective evaluation of recommenders and domain specific problems such as periodicity of recommendations, constrains, boosting, etc.

Follow Us

Copyright (c) Data Science Laboratory @ FIT CTU 2014–2016. All rights reserved.