Fusing Transformer Model with Temporal Features for ECG Heartbeat Classification

2019 ◽  
Author(s):  
Genshen Yan ◽  
Shen Liang ◽  
Yanchun Zhang ◽  
Fan Liu
Keyword(s):  
Heartbeat Classification ◽  
Temporal Features ◽  
Transformer Model

Enhancing dynamic ECG heartbeat classification with lightweight transformer model

2022 ◽  
pp. 102236
Author(s):  
Lingxiao Meng ◽  
Wenjun Tan ◽  
Jiangang Ma ◽  
Ruofei Wang ◽  
Xiaoxia Yin ◽  
...  
Keyword(s):  
Heartbeat Classification ◽  
Transformer Model

Robust wave-feature adaptive heartbeat classification based on self-attention mechanism using a transformer model

2021 ◽  
Author(s):  
Shuaicong Hu ◽  
Wenjie Cai ◽  
Tijie Gao ◽  
Jiajun Zhou ◽  
Mingjie Wang
Keyword(s):  
Model Performance ◽  
Attention Mechanism ◽  
Classification Method ◽  
Ectopic Beat ◽  
Supraventricular Arrhythmia ◽  
Segmentation Method ◽  
Ventricular Ectopic Beat ◽  
Heartbeat Classification ◽  
Symbolic Representations ◽  
Transformer Model

Abstract Objective: Electrocardiography is a common method for screening cardiovascular diseases. Accurate heartbeat classification assists in diagnosis and has attracted great attention. In this paper, we proposed an automatic heartbeat classification method based on a transformer neural network using a self-attention mechanism. Approach: An adaptive heartbeat segmentation method was designed to selectively focus on the time-dependent representation of heartbeats. A one-dimensional convolution layer was used to embed wave characteristics into symbolic representations, and then, a transformer block using multi-head attention was applied to deal with the dependence of wave-embedding. The model was trained and evaluated using the MIT-BIH arrhythmia database (MIT-DB). To improve the model performance, the model pre-trained on MIT-BIH supraventricular arrhythmia database (MIT-SVDB) was used and fine-tuned on MIT-DB. Main results: The proposed method was verified using the MIT-DB for two groups. In the first group, our method attained F1 scores of 0.86 and 0.96 for the supraventricular ectopic beat (SVEB) class and ventricular ectopic beat (VEB) class, respectively. In the second group, our method achieved an average F1 value of 99.83% and better results than other state-of-the-art methods. Significance: We proposed a novel heartbeat classification method based on a transformer model. This method provides a new solution for real-time electrocardiogram heartbeat classification, which can be applied to wearable devices.

Temporal Characteristics of Speech in Simultaneous Communication

1995 ◽  
Vol 38 (5) ◽  
pp. 1014-1024 ◽  
Author(s):  
Robert L. Whitehead ◽  
Nicholas Schiavetti ◽  
Brenda H. Whitehead ◽  
Dale Evan Metz
Keyword(s):  
Pause Duration ◽  
Vowel Duration ◽  
Vowel Perception ◽  
Temporal Characteristics ◽  
Spoken English ◽  
Temporal Features ◽  
Before And After ◽  
Word Duration ◽  
Rate Of Speech ◽  
Speech Condition

The purpose of this investigation was twofold: (a) to determine if there are changes in specific temporal characteristics of speech that occur during simultaneous communication, and (b) to determine if known temporal rules of spoken English are disrupted during simultaneous communication. Ten speakers uttered sentences consisting of a carrier phrase and experimental CVC words under conditions of: (a) speech, (b) speech combined with signed English, and (c) speech combined with signed English for every word except the CVC word that was fingerspelled. The temporal features investigated included: (a) sentence duration, (b) experimental CVC word duration, (c) vowel duration in experimental CVC words, (d) pause duration before and after experimental CVC words, and (e) consonantal effects on vowel duration. Results indicated that for all durational measures, the speech/sign/fingerspelling condition was longest, followed by the speech/sign condition, with the speech condition being shortest. It was also found that for all three speaking conditions, vowels were longer in duration when preceding voiced consonants than vowels preceding their voiceless cognates, and that a low vowel was longer in duration than a high vowel. These findings indicate that speakers consistently reduced their rate of speech when using simultaneous communication, but did not violate these specific temporal rules of English important for consonant and vowel perception.

Power Losses in RF Inductor of Ferrite-Free Closed-Loop Inductively-Coupled Low Pressure Mercury Lamps

2020 ◽  
pp. 89-94 ◽  
Author(s):  
Ekaterina V. Lovlya ◽  
Oleg A. Popov
Keyword(s):  
Closed Loop ◽  
Low Pressure ◽  
Mercury Vapour ◽  
Plasma Power ◽  
Power Losses ◽  
Inductive Discharge ◽  
Inductively Coupled ◽  
Radial Inhomogeneity ◽  
Transformer Model ◽  
Rf Inductor

RF inductor power losses of ferrite-free electrode-less low pressure mercury inductively-coupled discharges excited in closed-loop dielectric tube were studied. The modelling was made within the framework of low pressure inductive discharge transformer model for discharge lamps with tubes of 16, 25 and 38 mm inner diam. filled with the mixture of mercury vapour (7.5×10–3 mm Hg) and argon (0.1, 0.3 and 1.0 mm Hg) at RF frequencies of 1, 7; 3.4 and 5.1 MHz and plasma power of (25–500) W. Discharges were excited with the help of the induction coil of 3, 4 and 6 turns placed along the inner perimeter of the closed-loop tube. It was found that the dependence of coil power losses, Pcoil, on the discharge plasma power, Ppl, had the minimum while Pcoil decreased with RF frequency, tube diameter and coil number of turns. The modelling results were found in good qualitative agreement with the experimental data; quantitative discrepancies are believed to be due skin-effect and RF electric field radial inhomogeneity that were not included in discharge modelling.

Der Begriff ‚Anlage‘ als Bestimmung des Seins bei Martin Heidegger

2010 ◽  
Vol 2010 (1) ◽  
pp. 101-116
Author(s):  
Ales Novak
Keyword(s):  
Martin Heidegger ◽  
20Th Century ◽  
Being And Time ◽  
Temporal Features ◽  
Heaven And Earth

During the philosophical pathway of Martin Heidegger the 30s of the 20th century are a crucial period in respect of his effort to point out the temporal meaning of the notion of being. After the failure of his project of Being and Time he turned his attention towards pondering upon the (Hi)Story of being (Seinsgeschichte or Geschichte des Seins), leading him to the thought of the oblivion of being as well as of the forsakenness by the being. Within the eschatological perspectives after the end of metaphysics Heidegger arrives at the notion of Anlage, in which he means to articulate the temporal features of being corresponding to the mentioned epochal situation. The notion Anlage sums up the temporal features of setting, perpetuity, and presence, which according to Heidegger are notoriously associated with the notion of being within the metaphysics. Nonetheless, even this conceptual effort acts as a taking- off towards a far more radical phenomenology of world conceived as the fourfold of heaven and earth, the divine and the mortals.

scholarly journals A Single-Trial P300 Detector Based on Symbolized EEG and Autoencoded-(1D)CNN to Improve ITR Performance in BCIs

Sensors ◽  
2021 ◽  
Vol 21 (12) ◽  
pp. 3961
Author(s):  
Daniela De Venuto ◽  
Giovanni Mezzina
Keyword(s):  
Recognition Performance ◽  
Low Complexity ◽  
Single Trial ◽  
Eeg Signals ◽  
Speed Increase ◽  
P300 Speller ◽  
Embedded Platform ◽  
Temporal Features ◽  
High Recognition Accuracy ◽  
Car Driving

In this paper, we propose a breakthrough single-trial P300 detector that maximizes the information translate rate (ITR) of the brain–computer interface (BCI), keeping high recognition accuracy performance. The architecture, designed to improve the portability of the algorithm, demonstrated full implementability on a dedicated embedded platform. The proposed P300 detector is based on the combination of a novel pre-processing stage based on the EEG signals symbolization and an autoencoded convolutional neural network (CNN). The proposed system acquires data from only six EEG channels; thus, it treats them with a low-complexity preprocessing stage including baseline correction, windsorizing and symbolization. The symbolized EEG signals are then sent to an autoencoder model to emphasize those temporal features that can be meaningful for the following CNN stage. This latter consists of a seven-layer CNN, including a 1D convolutional layer and three dense ones. Two datasets have been analyzed to assess the algorithm performance: one from a P300 speller application in BCI competition III data and one from self-collected data during a fluid prototype car driving experiment. Experimental results on the P300 speller dataset showed that the proposed method achieves an average ITR (on two subjects) of 16.83 bits/min, outperforming by +5.75 bits/min the state-of-the-art for this parameter. Jointly with the speed increase, the recognition performance returned disruptive results in terms of the harmonic mean of precision and recall (F1-Score), which achieve 51.78 ± 6.24%. The same method used in the prototype car driving led to an ITR of ~33 bit/min with an F1-Score of 70.00% in a single-trial P300 detection context, allowing fluid usage of the BCI for driving purposes. The realized network has been validated on an STM32L4 microcontroller target, for complexity and implementation assessment. The implementation showed an overall resource occupation of 5.57% of the total available ROM, ~3% of the available RAM, requiring less than 3.5 ms to provide the classification outcome.

Sign in / Sign up

Export Citation Format

Share Document