Browsing by Author "Leung, Henry"

Now showing 1 - 20 of 59
  • Thumbnail Image
    ItemOpen Access
    A novel chaos based UWB radar and its application to through-the-wall sensing
    (2005) Venkatasubramanian, Vijayaraghavan; Leung, Henry
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    A Novel Feature-Level Data Fusion Method for Indoor Autonomous Localization
    (2013-07-14) Liu, Minxiang; Wang, Yuhao; Leung, Henry; Yu, Jiangnan
    We present a novel feature-level data fusion method for autonomous localization in an inactive multiple reference unknown indoor environment. Since monocular sensors cannot provide the depth information directly, the proposed method incorporates the edge information of images from a camera with homologous depth information received from an infrared sensor. Real-time experimental results demonstrate that the accuracies of position and orientation are greatly improved by using the proposed fusion method in an unknown complex indoor environment. Compared to monocular localization, the proposed method is found to have up to 70 percent improvement in accuracy.
  • Thumbnail Image
    ItemOpen Access
    A Novel Ranking Method Based on Subjective Probability Theory for Evolutionary Multiobjective Optimization
    (2011-09-15) Wei, Shuang; Leung, Henry
    Most of the engineering problems are modeled as evolutionary multiobjective optimization problems, but they always ask for only one best solution, not a set of Pareto optimal solutions. The decision maker's subjective information plays an important role in choosing the best solution from several Pareto optimal solutions. Generally, the decision-making processing is implemented after Pareto optimality. In this paper, we attempted to incorporate the decider's subjective sense with Pareto optimality for chromosomes ranking. A new ranking method based on subjective probability theory was thus proposed in order to explore and comprehend the true nature of the chromosomes on the Pareto optimal front. The properties of the ranking rule were proven, and its transitivity was presented as well. Simulation results compared the performance of the proposed ranking approach with the Pareto-based ranking method for two multiobjective optimization cases, which demonstrated the effectiveness of the new ranking approach.
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    An Adaptive Kernel Layer Deep Neural Network for Remote Sensing Imageries
    (2022-08-02) Al Shoura, Tariq; Leung, Henry; Bento, Mariana; Ioannou, Yani
    Convolutional neural networks in recent years have become wildly utilized for the various applications that deal with images, going through different architectures and development to improve their capabilities for features extraction in order to yield better results in deep learning methods. Since images are vital resources that are used to represent various sceneries; multi-temporal images are an important tool used to monitor the changes that happen to those sceneries, which is required for multiple fields such as urban planning and disaster assessment, and with the rapid development of technology, very high-resolution (VHR) images from various sources are now more available, requiring analysis of the convolutional networks for larger images, and further developments to enhance their performance, so that they can process the large amount of data and identify any changes more efficiently. This thesis proposes the utilization of adaptive kernels layer (AKL) in order to extract features from large images, where the layer has been designed to maximize spectral information while retaining the spatial resolution of the information, discussing the benefits gained from utilizing it, and comparing it to other popular feature selection methods in image change detection applications. This thesis also examines and provides models used to generate the change image by utilizing machine learning (ML) models, mainly convolutional neural networks (CNN) and long shortterm memory (LSTM).
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    An Evaluation of Chaos Modulation In Wireless Acoustic Sensor Networks
    (2014-05-14) Oluge, Dasola; Leung, Henry
    Wireless sensor networks are deployed in different regions of interest, and used in many applications such as surveillance and telemedicine. Advancements in wireless communication technology and miniaturization of electronic devices have led to sensors nodes which can be deployed in different terrains such as underwater with acoustic sensing functionality. However, non-ideal channel conditions in such terrains can affect the sensor network performance. In this thesis, the chaos modulation schemes are shown to perform better than conventional Direct Sequence Spread Spectrum in propagation delay channels because they employ non-coherent demodulators. Furthermore, since propagation delay is rampant in the underwater environment, an integrated simulation framework is used to evaluate the network level performance with choice of modulation scheme in the underwater environment. Ergodic Chaos Parameter Modulation scheme performs best and thus proposed for use in underwater acoustic localization using wireless acoustic sensor networks.
  • Thumbnail Image
    ItemOpen Access
    An insect like navigation mobile robot based on lyapunov control and fuzzy logic technique
    (2012) Lei, Ting; Macnab, Chris; Leung, Henry
    A novel reactive navigation strategy is tested in this thesis, in both simulation and experiment. Previous methods in the literature mostly use path planning strategies to avoid all obstacles in the environment, assuming all obstacles can be reliably located and classified. An alternative strategy relies on sensing only immediate obstacles in the path and avoiding them using a reactive strategy-similar to how an insect avoids obstacles in a natural environment. A reactive strategy is investigated in this thesis, consisting of three basic components: 1) a fuzzy-logic algorithm for choosing the heading based on a target location and perceived immediate-obstacle directions, 2) a nonlinear closed-loop control system that commands appropriate angular velocities for achieving the heading, and 3) a fuzzy obstacle ahead truth value term added directly to commanded angular and translational velocities which immediately changes robot direction to avoid imminent collisions. The proposed strategy is tested in a simulation of a maze-like environment, and the effect of all controller parameters is evaluated through extensive testing. An experimental mobile robot, using a laser range finder to detect obstacles and GPS to determine ultimate target locations, uses the algorithm to navigate around a tree in an outdoor environment.
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    Analyzing Causality between Actual Stock Prices and User-weighted Sentiment in Social Media for Stock Market Prediction
    (2016) Park, Jin-Tak; Leung, Henry; Far, Behrouz; Ruhe, Guenther
    In this thesis, an improved sentiment analysis algorithm is proposed which reflects the impact of user, and to analyze whether public sentiment calculated by the proposed algorithm can contribute to stock prediction. The proposed sentiment analysis algorithm reflects the factors of Twitter which are relevant to users’ authority to calculate sentiment weight of each message that is different from existing sentiment analysis algorithms. Linear and nonlinear prediction models are constructed to forecast future stock prices of selected companies. The proposed algorithm is applied to both linear and nonlinear prediction models and comparisons of prediction accuracy with the existing sentiment analysis algorithm are performed. To support the approach of the proposed algorithm that the authoritative users affect the other users, causal relationship between them is figured out through Granger Causality analysis. Further analysis is also provided to find causal relationship between public sentiment and the actual changes of the stock prices.
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    Bayesian Sparse Estimation Using Double Lomax Priors
    (2013-08-27) Gu, Xiaojing; Leung, Henry; Gu, Xingsheng
    Sparsity-promoting prior along with Bayesian inference is an effective approach in solving sparse linear models (SLMs). In this paper, we first introduce a new sparsity-promoting prior coined as Double Lomax prior, which corresponds to a three-level hierarchical model, and then we derive a full variational Bayesian (VB) inference procedure. When noninformative hyperprior is assumed, we further show that the proposed method has one more latent variable than the canonical automatic relevance determination (ARD). This variable has a smoothing effect on the solution trajectories, thus providing improved convergence performance. The effectiveness of the proposed method is demonstrated by numerical simulations including autoregressive (AR) model identification and compressive sensing (CS) problems.
  • Thumbnail Image
    ItemOpen Access
    Bio-Inspired Collective Decision Making in Multiagent System: From Low- to High-Cognition Algorithms
    (2014-05-14) Aichour, Hichem Zakaria; Leung, Henry
    At the beginning, this thesis proposes an algorithm inspired by ant's scout-and-recruit for foraging providing a low-cognition algorithm to rescue civilians in distress and explaining how it could be achieved. Being low-cognition algorithm, this algorithm is characterized by having minimal computation and communication between the agents of the MAS as well as being applicable for large scale environment due to its inspiration from ants. Then, an algorithm inspired by human's conformity is provided which is considered a high-cognition algorithm. This algorithm is applied to a rescue mission that is simulated using \emph{RoboCup Rescue Simulator}. Being high-cognition algorithm, this algorithm is characterized by having heavy computation and communication made by the agents of the MAS. Finally, the conformity algorithm is integrated on top of the scout-and-recruit algorithm to enhance its performance. While the low-cognition part insures its applicability to large scale, the high-cognition algorithm pushes the agents to make smarter decisions enhancing the overall system performance.
  • Thumbnail Image
    ItemOpen Access
    A Biologically Inspired Supervised Learning Rule for Audio Classification with Spiking Neural Networks
    (2021-06-15) Peterson, Dylan George; Leung, Henry; Westwick, David; Uddin, Gias; Wang, Xin
    Audio classification has many practical applications such as noise pollution monitoring, wildlife monitoring, audio surveillance, speech recognition, and more. For many of these applications, deploying classifiers on low powered devices for persistent monitoring is desirable. Artificial neural networks (ANN) have achieved significant success for audio classification tasks. However, it may not always be feasible to deploy current state-of-the-art ANNs to embedded devices due to their memory footprint and power consumption. Biologically inspired neural networks, also known as spiking neural networks (SNN), have been shown to significantly reduce power consumption during inference when compared with equivalent ANNs. They have also been theoretically proven to be more computationally powerful per unit than ANNs. These two properties make SNNs an attractive solution for machine learning tasks on low powered embedded devices, such as at the edge in an Internet of Things (IoT) sensor network. However, SNNs tend to lag behind in performance when compared to ANNs. This is partially because training SNNs is difficult since the standard backpropagation algorithm is not directly applicable due to the non-differentiable spiking nature of SNNs. Encoding data into spike trains compatible with SNNs is also an unresolved question when applying SNNs. This work compares different spike encoding schemes for audio data, and a learning algorithm for multilayer SNNs inspired by biologically plausible learning rules is developed. The proposed learning rule is then successfully applied to simple pattern recognition and audio classification tasks.
  • Thumbnail Image
    ItemOpen Access
    Chaos Modulation and Equalization for Robust Wireless Communications
    (2022-01-28) Li, Boyuan; Leung, Henry; Messier, Geoffrey; Helaoui, Mohamed; Fapojuwo, Abraham; Kaddoum, Georges
    This thesis focuses on using chaos modulation and equalization to enhance the robustness of wireless communication. Our contributions are three-fold: For the Industrial Internet of Things (IIoT), a quadratic ergodic chaotic parameter modulation (QECPM) is proposed. We use software-defined radios (SDRs) to show that QECPM is robust against timing synchronization errors, which have a major effect on performance. The bit-error-rate (BER) performance of QECPM in Nakagami-m fading channels is derived and verified by simulations. In a multipath-rich channel, QECPM demonstrates superior performance to conventional modulations. Furthermore, we show that retransmissions causes misaligned packets; however, when using the proposed receiver, the error bits are sparse enough to utilize the non-retransmission mode to maintain stable link rates. In a denoise-and-forward (DNNF) two-way relay system, the signals are asynchronous. For most denoising and decoding methods, precise estimation of the delay is required by the relay and end users, which is not always available. Using the ergodic property of chaotic signals, we propose to address the asynchronous problem using ergodic chaotic parameter modulation (ECPM) and guarding intervals (GIs). The theoretical BER performance is analyzed and verified by simulations. A relay selection method is also proposed for two-way relay systems with multiple relays to achieve improved performance compared to using all relays. Chaos modulation signals can be blindly equalized using phase space volume (PSV). A maximum likelihood-PSV (ML-PSV) estimation and the Cramer Rao Lower Bound (CRLB) are derived. The ML-PSV algorithm is applied to blind system identification of autoregressive (AR) and moving average (MA) models as well as equalization. A method for ECPM to identify the constructive/destructive channel is developed. The destructive channel effect can be mitigated using the proposed equalization. Our approach is validated using SDRs. Our results show that ECPM with ML-PSV equalization is more robust than comparing methods.
  • Thumbnail Image
    ItemOpen Access
    Chaotic data hiding for multimedia
    (2005) Chen, Siyue; Leung, Henry
  • Thumbnail Image
    ItemOpen Access
  • Thumbnail Image
    ItemOpen Access
    Characterization and Modeling of Brain Tissues Using Fractional Calculus
    (2018-02) Jiao, Shanlin; Sun, Qiao; Sutherland, Garnette; Goldsmith, Peter; Leung, Henry; Pieper, Jeff K.
    Modeling of brain tissues is essential to better patient outcome. This research aims at modeling brain tissues based on their electrical impedance and viscoelasticity. Fractional order models can accurately model the two properties with few parameters in a wide spectral range. The Cole parameters extracted from step current response was applied to characterize the grey and white matter. Experimental results show that the Cole model fits well to experimental data and proposed Cole parameter extraction method is more effective in identifying Cole parameters. A fractional order viscoelastic model is employed to model the viscoelasticity of brain tissue. The predicted results are compared with the known experimental data and also that of integer order models, indicating the fractional order viscoelastic model can adequately fit all the experimental data with only two parameters.