Running projects

  • Picture of General Game Playing

    General Game Playing

    This research project aims towards optimization of general game controllers or more generally solving problems of reinforcement learning. Currently we experiment with games on Atari 2600 platform. The goal is to develop a method able to fully-automatically generate a general controler for a chosen set of games. The contoller takes unpreprocessed game screen images as its input. The output corresponds with joystick movement. We apply state-of-the-art approaches to solve the problem: deep networks (convolutional neural networks), indirect encodings of neural networks, evolutionary algorithms and advanced reinforcement learning techniques. The outcomes of this project might form a stepping stone towards general AI.

    More »

  • Picture of FAKE GAME

    FAKE GAME

    We are developing the open source software FAKE GAME. This software should be able to automatically preprocess various data, to generate regressive, predictive models and classifiers (by means of GAME engine), to automatically identify interesting relationships in data (even in high-dimensional ones) and to present discovered knowledge in a comprehensible form. The software should fill gaps which are not covered by existing open source data mining environment WEKA and possibly integrate with the YALE environment.

    More »

  • Picture of Java Continuous Optimization Library

    Java Continuous Optimization Library (JCool)

    JCool is an open-source Java library for continuous optimization. It implements both numerical (Gradient Descent, Conjugated Gradients, Quasi-Newton,..) and nature-inspired (Evolutionary, PSO, ACO,.. ) algorithms. We have included a huge set of benchmarking problems and a tool which compares methods statistically.

    More »

Completed projects

  • Picture of Computational Intelligence and Voice Processing Toolkit for Cell Broadband Engine

    Computational Intelligence and Voice Processing Toolkit for Cell Broadband Engine (CIV Toolkit)

    The goal of the proposed project is to develop software toolkit (library) for the Cell Broadband Engine (CBE). The toolkit will provide a collection of advanced algorithms belonging to the computational intelligence and voice processing areas. The toolkit is developed as an open source project on the top of CBE SDK libraries. This project runs in close cooperation with specialists from IBM Czech Republic and it is supported by IBM Faculty Awards program.

    More »

  • Picture of BlueCar

    BlueCar

    BlueCar is a mobile robot with remote BlueTooth control. This robot caries bidirectional audio port (microphone and speaker) which is connected over BlueTooth to a remote computer with IBM speach recognition technology installed. This robot is intended as a helper for intelligent rooms.

    More »

  • Picture of Bone Age Modelling

    Bone Age Modelling

    In this project, we focus on processing Anthropological data by means of several data mining methods. The goal is to predict an age of individuals described by a set of parameters measured on their skeletons. Data in this project are problematic due to very high noise. Methods are tuned and parameterized to give best possible performance on data. The performance of methods is compared and the recommendation, how to process noisy and partially inconsistent data will be one of the final conclusions of this project.

    More »

  • Class Assignment Tool (CATool)

    The CATool is an application which supports data preprocessing - mainly signals. It allows to import of data in different formats. More it allows many different types of transformations - fft, wavelet, statistics, signal filtering, etc... And support export to many file formats suitable for data mining (Weka Arff, Yale, GAME Input format, plain text, ... ). More it allows you to score (assign classes) to each segment of the signal.

    More »

  • Picture of Colabroute

    Colabroute

    Colabroute projects focuses on acquisition, building and knowledge extraction from GPS captured data. Data collected from vehicles contain a lot of features that can be exploited for route planning, dynamic routing and navigation. Maps are obtained automatically without any external map source. With Colabroute, users can map any unexplored area on Earth.

    More »

  • EnhancedDOP

    A processor with neural network support is designed within this project. Project is based on the DOP procoessor that was developed by VLSI group of our department. The instruction set of this processor is enhanced by a set of neural instructions that implement operations typical for neural networks (activation function, dot product, ...). The ALU consists of an adder and barrel shifter, no hardware multiplier is necessary to implement because imprecise shift-add arithmetics is used. This processor is intended for embedded applications.

    More »

  • Hardware Accelerated Computational Inteligence (HACI)

    The HACI project aim is to implement adaptive algorithms in hardware, exploit their natural paralelism and develop an implementation of adaptive algorithms retaining the precision and error close to the original software implementation.

    More »

  • Picture of Neuroevolution of Robots Behavior

    Neuroevolution of Robots Behavior (robo)

    The aim of the project is to evolve artificial neural networks as controllers for agents (simulated robots) that complete various tasks. Neural networks are evolved by HyperNEAT (Stanley). Robots are simulated using Vivae environment. By now robots are able to stay and drive on roads instead of grass surface with a high friction. Other experiments deal with obstacle avoidance. We work also on HyperGP which is HyperNEAT with NEAT substituted by Genetic Programming.

    More »

  • Parkinson Disease Diagnostics (PDD)

    The goal of our project is to support disgnostic of patients with Parkinson disease (and other movement disorders). Patients performs proposed tests on tablet, data are collected and stored in database. Our application provides set of viewers and analytic tools for tremor assesment.

    More »

  • Picture of Parrot Speech

    Parrot Speech

    Parrot Speech project deals with an implementation of a system that helps to analyze vocalization of gray parrots. The system extract acoustic features from a real recorded phonic samples. The samples are processed using various speech signal processing methods and finally analyzed by self-organizing neural networks.

    More »

  • Picture of ShapeSOM

    ShapeSOM

    The aim of this project is to exploit a power of self-organization for intelligent reconstruction of 3-D shapes from laser range finder. We apply artificial neural networks that moves the task to three dimensions.

    More »

  • Picture of Simulator of Modular Neural Networks

    Simulator of Modular Neural Networks (SiMoNNe)

    SiMoNNe is a modular neural network simulator. It is the language driven machine. Simulator is not independent on praticular neural paradigm which means that the neural networks with variable topology and function can be simulated using SiMoNNe. The communication with simulator is based on a simple language which can describe the network, control experiment and return result. The results of the simulation are written in the SiMoNNe language too and are reusable as a simulator input. The SiMoNNe simulator is written in pure Java so it is platform independent. The simulator is determined to simulate modular neural networks.

    More »

  • StockSOM

    StockSOM is a SOM based application for support of stock trading. It utilizes mining and clustering of text information stored in short headlines from MarketWatch server.

    More »

  • Picture of Temporal Information Categorizing and Learning Map

    Temporal Information Categorizing and Learning Map (TICALM)

    TICALM is an advanced recurrent self-organizing neural network which is capable of a multidimensional signal recognition. TICALM is based on enhanced version of Categorizing and Learnin Map (CALM) developed for static pattern categorization. TICALM embeds creates spatial and temporal model of an input signal (temporal sequence of spatial patterns) that can be exploited in signal modelling recognition and prediction.

    More »

  • Picture of ZISC Online

    ZISC Online

    The ZISC Online project utilizes a hardware artificial neural network card based on IBM ZISC036 neurochips. The ZISC036 simulates RBF network neurons. The project aims to develop server driver for the card and ease an access to the card via Internet.

    More »