We exploit altered patterns in brain functional connectivity as features for automatic discriminative analysis of neuropsychiatric patients. Deep learning methods have been introduced to network classification only very recently fMRI, and the proposed architectures essentially focused on a single type measure. propose deep convolutional neural (CNN) framework electroencephalogram (EEG)-derived connectome schizophrenia (SZ). To capture complementary aspects disrupted SZ, we explore...

We propose a novel Attentional Scale Sequence Fusion based You Only Look Once (YOLO) framework (ASF-YOLO) which combines spatial and scale features for accurate fast cell instance segmentation. Built on the YOLO segmentation framework, we employ Feature (SSFF) module to enhance multiscale information extraction capability of network, Triple Encoder (TFE) fuse feature maps different scales increase detailed information. further introduce Channel Position Attention Mechanism (CPAM) integrate...

Brain functional connectivity (FC) networks inferred from magnetic resonance imaging (fMRI) have shown altered or aberrant brain connectome in various neuropsychiatric disorders. Recent application of deep neural to connectome-based classification mostly relies on traditional convolutional (CNNs) using input FCs a regular Euclidean grid learn spatial maps neglecting the topological information networks, leading potentially sub-optimal performance disorder identification. We propose novel...

This paper proposes a framework based on deep convolutional neural networks (CNNs) for automatic heart sound classification using short-segments of individual beats. We design 1D-CNN that directly learns features from raw heart-sound signals, and 2D-CNN takes inputs two- dimensional time-frequency feature maps Mel-frequency cepstral coefficients (MFCC). further develop CNN ensemble (TF-ECNN) combining the score-level fusion class probabilities. On large PhysioNet CinC challenge 2016...

We consider the challenges in estimating state-related changes brain connectivity networks with a large number of nodes. Existing studies use sliding-window analysis or time-varying coefficient models, which are unable to capture both smooth and abrupt simultaneously, rely on ad-hoc approaches high-dimensional estimation. To overcome these limitations, we propose Markov-switching dynamic factor model, allows states functional magnetic resonance imaging (fMRI) data be driven by...

Objective: This paper considers challenges in developing algorithms for accurate segmentation and classification of heart sound (HS) signals. Methods: We propose an approach based on Markov switching autoregressive model (MSAR) to segmenting the HS into four fundamental components each with distinct second-order structure. The identified boundaries are then utilized automated pathological using continuous density hidden (CD-HMM). MSAR formulated a state-space form is able capture...

Objective: This paper addresses the critical problem of estimating time-evolving effective brain connectivity. Current approaches based on sliding window analysis or time-varying coefficient models do not simultaneously capture both slow and abrupt changes in causal interactions between different regions. Methods: To overcome these limitations, we develop a unified framework switching vector autoregressive (SVAR) model. Here, dynamic connectivity regimes are uniquely characterized by...

To study the effective connectivity among sources in a densely voxelated (high-dimensional) cortical surface, we develop source-space factor VAR model. The first step our procedure is to estimate activity from multichannel electroencephalograms (EEG) using anatomically constrained brain imaging methods. Following parcellation of surface into disjoint regions interest (ROIs), latent factors within each ROI are computed principal component analysis. These ROI-specific low-rank approximations...

Soft sensing of hard-to-measure variables is often crucial in industrial processes. Current practices rely heavily on conventional modeling techniques that show success improving accuracy. However, they overlook the non-linear nature, dynamics characteristics, and non-Euclidean dependencies between complex process variables. To tackle these challenges, we present a framework known as Knowledge discovery graph Attention Network for effective (KANS). Unlike existing deep learning soft sensor...

Higher-order sensor networks are more accurate in characterizing the nonlinear dynamics of sensory time-series data modern industrial settings by allowing multi-node connections beyond simple pairwise graph edges. In light this, we propose a deep spatio-temporal hypergraph convolutional neural network for soft sensing (ST-HCSS). particular, our proposed framework is able to construct and leverage higher-order (hypergraph) model complex multi-interactions between nodes absence prior...

Disrupted functional connectivity patterns have been increasingly used as features in pattern recognition algorithms to discriminate neuropsychiatric patients from healthy subjects. Deep neural networks (DNNs) were employed fMRI network classification only very recently and its application EEG-based connectome is largely unexplored. We propose a DNN with deep belief (DBN) architecture for automated of schizophrenia (SZ) based on EEG effective connectivity. vector-autoregression-based...

We present a unified statistical framework for characterizing community structure of brain functional networks that captures variation across individuals and evolution over time. Existing methods detection focus only on single-subject analysis dynamic networks; while recent extensions to multiple-subjects are limited static networks. To overcome these limitations, we propose multi-subject, Markov-switching stochastic block model (MSS-SBM) identify state-related changes in organization group...

Malay speech therapy assistance tools (MSTAT) is a system which assists the therapist to diagnose children for language disorder and train with stuttering problem. The main engine behind it technologies; consist of recognition system, text-to-speech talking head by Tan, T.S. (2003). In this project, utilizes hidden Markov model (HMM) technique evaluating problem such as stuttering. voice pattern normal are used HMM classifying disorder. Thus, localized local dialect database focus on that...

Problem statement: In facial biometrics, face features are used as the required human traits for automatic recognition. Feature extracted from images significant biometrics system performance. Approach: this thesis, a framework of biometric was designed based on two subspace methods i.e., Principal Component Analysis (PCA) and Linear Discriminant (LDA). First, PCA is dimension reduction, where original projected into lower-dimensional representations. Second, LDA proposed to provide solution...

This paper proposes a new approach for electrocardiogram (ECG) based personal identification on extended Kalman filtering (EKF) framework. The framework uses nonlinear ECG dynamic models formulated to represent noisy signal. advantage of the is ability capture distinct features used biometric recognition such as temporal and amplitude distances between PQRST points. Moreover inherent modeling additive noise provides robust recognition. Log-likelihood scoring proposed classification....