Ústav počítačových systémů


Recent Submissions

Now showing 1 - 5 of 7
  • Item
    Deep learning-based assessment model for Real-time identification of visual learners using Raw EEG
    (2024-01-02) Jawed, Soyiba; Faye, Ibrahima; Malik, Aamir Saeed
    Automatic identification of visual learning style in real time using raw electroencephalogram (EEG) is challenging. In this work, inspired by the powerful abilities of deep learning techniques, deep learning-based models are proposed to learn high-level feature representation for EEG visual learning identification. Existing computer-aided systems that use electroencephalograms and machine learning can reasonably assess learning styles. Despite their potential, offline processing is often necessary to eliminate artifacts and extract features, making these methods unsuitable for real-time applications. The dataset was chosen with 34 healthy subjects to measure their EEG signals during resting states (eyes open and eyes closed) and while performing learning tasks. The subjects displayed no prior knowledge of the animated educational content presented in video format. The paper presents an analysis of EEG signals measured during a resting state with closed eyes using three deep learning techniques: Long-term, short-term memory (LSTM), Long-term, short-term memory-convolutional neural network (LSTM-CNN), and Long-term, short-term memory - Fully convolutional neural network (LSTM-FCNN). The chosen techniques were based on their suitability for real-time applications with varying data lengths and the need for less computational time. The optimization of hypertuning parameters has enabled the identification of visual learners through the implementation of three techniques. LSTM- CNN technique has the highest average accuracy of 94%, a sensitivity of 80%, a specificity of 92%, and an F1 score of 94% when identifying the visual learning style of the student out of all three techniques. This research has shown that the most effective method is the deep learning-based LSTM-CNN technique, which accurately identifies a student's visual learning style.
  • Item
    Formal Methods for Exact Analysis of Approximate Circuits
    (2019-12-21) Vašíček, Zdeněk
    Approximate circuits are digital circuits that are intentionally designed in such a way that the specification is violated in terms of functionality in order to obtain some improvements in power consumption, performance or area, in comparison with fully functional circuits.  To design the approximate circuits, the synthesis tools rely on the availability of a procedure checking, whether the synthesized circuits meet a specification and/or provides information about circuit quality. Compared to the traditional circuit design flow, the nature of the approximate circuits involves replacing the strict functional equivalence checking with a more advanced approach that enables us to quantify or guarantee the degree of similarity. The most common technique is to employ a circuit simulator for analysing responses for all input vectors. This approach allows us to simultaneously perform checking and quality assessment, but the exhaustive enumeration of the input vectors is tractable only for a small number of inputs. To avoid excessive run-times, a subset of all possible input vectors is typically used for complex circuits. This causes us, however, to lose the ability to guarantee that the quality of the synthesized circuits is within an acceptable range given in the specification.  The main goal of this paper is to show how to adopt formal methods such as binary decision diagrams and satisfiability solvers for exhaustive analysis of approximate circuits without explicit enumeration of all input vectors. We survey the methods for exact computation of the most important error parameters used in the context of approximate computing, propose improved algorithms and provide a detailed analysis of their performance.  The methods are benchmarked on a large set of key approximate circuits consisting of nearly 2,000 unique arithmetic instances with 8-, 12-, 16-, and 32-bit operands which helps us to identify the best algorithm and method for computation of a desired error parameter.
  • Item
    Efficient Low-Resource Compression of HIFU Data
    (2018-06-26) Klepárník, Petr; Bařina, David; Zemčík, Pavel; Jaroš, Jiří
    Large-scale numerical simulations of high-intensity focused ultrasound (HIFU), important for model-based treatment planning, generate large amounts of data. Typically, it is necessary to save hundreds of gigabytes during simulation. We propose a novel algorithm for time-varying simulation data compression specialised for HIFU. Our approach is particularly focused on on-the-fly parallel data compression during simulations. The algorithm is able to compress 3D pressure time series of linear and non-linear simulations with very acceptable compression ratios and errors (over 80% of the space can be saved with an acceptable error). The proposed compression enables significant reduction of resources, such as storage space, network bandwidth, CPU time, and so forth, enabling better treatment planning using fast volume data visualisations. The paper describes the proposed method, its experimental evaluation, and comparisons to the state of the arts.
  • Item
    Accurate simulation of transcranial ultrasound propagation for ultrasonic neuromodulation and stimulation
    (2017-03-13) Robertson, James; Cox, Ben; Jaroš, Jiří; Treeby, Bradley
    Non-invasive, focal neurostimulation with ultrasound is a potentially powerful neuroscientific tool that requires effective transcranial focusing of ultrasound to develop. Time-reversal (TR) focusing using numerical simulations of transcranial ultrasound propagation can correct for the effect of the skull, but relies on accurate simulations. Here, focusing requirements for ultrasonic neurostimulation are established through a review of previously employed ultrasonic parameters, and consideration of deep brain targets. The specific limitations of finite-difference time domain (FDTD) and k-space corrected pseudospectral time domain (PSTD) schemes are tested numerically to establish the spatial points per wavelength and temporal points per period needed to achieve the desired accuracy while minimizing the computational burden. These criteria are confirmed through convergence testing of a fully simulated TR protocol using a virtual skull. The k-space PSTD scheme performed as well as, or better than, the widely used FDTD scheme across all individual error tests and in the convergence of large scale models, recommending it for use in simulated TR. Staircasing was shown to be the most serious source of error. Convergence testing indicated that higher sampling is required to achieve fine control of the pressure amplitude at the target than is needed for accurate spatial targeting.
  • Item
    Effective EEG Feature Selection for Interpretable MDD (Major Depressive Disorder) Classification
    (Association for Computing Machinery, 2023-04-14) Mrázek, Vojtěch; Jawed, Soyiba; Arif, Muhammad; Malik, Aamir Saeed
    In this paper, we propose an interpretable electroencephalogram (EEG)-based solution for the diagnostics of major depressive disorder (MDD). The acquisition of EEG experimental data involved 32 MDD patients and 29 healthy controls. A feature matrix is constructed involving frequency decomposition of EEG data based on power spectrum density (PSD) using the Welch method. Those PSD features were selected, which were statistically significant. To improve interpretability, the best features are first selected from feature space via the non-dominated sorting genetic (NSGA-II) evolutionary algorithm. The best features are utilized for support vector machine (SVM), and k-nearest neighbors (k-NN) classifiers, and the results are then correlated with features to improve the interpretability. The results show that the features (gamma bands) extracted from the left temporal brain regions can distinguish MDD patients from control significantly. The proposed best solution by NSGA-II gives an average sensitivity of 93.3%, specificity of 93.4% and accuracy of 93.5%. The complete framework is published as open-source at https://github.com/ehw-fit/eeg-mdd.