ARTICLE | doi:10.20944/preprints202110.0111.v1
Subject: Mathematics & Computer Science, Artificial Intelligence & Robotics Keywords: iEEG; non-stationarity; lead seizure; seizure prediction; support vector machines; unbalanced classification; group learning
Online: 7 October 2021 (08:21:21 CEST)
We describe a novel system for online prediction of lead seizures from long-term intracranial electroencephalogram (iEEG) recordings for canines with naturally occurring epilepsy. This study adopts new specification of lead seizures, reflecting strong clustering of seizures in observed data. This clustering results in fewer lead seizures (~7 lead seizures per dog), and hence new challenges for online seizure prediction, that are addressed in the proposed system. In particular, machine learning part of the system is implemented using the Group Learning method suitable for modeling sparse and noisy seizure data. In addition, several modifications for the proposed system are introduced to cope with non-stationarity of noisy iEEG signal. They include: (1) periodic re-training of SVM classifier using most recent training data; (2) removing samples with noisy labels from training data; (3) introducing new adaptive post-processing technique for combining many predictions made for 20-second windows into a single prediction for 4 hr segment. Application of the proposed system requires only 2 lead seizures for training the initial model, and results in high prediction performance for all four dogs (with mean 0.84 sensitivity, 0.27 time-in-warning, and 0.78 false-positive rate per day). Proposed system achieves accurate prediction of lead seizures during long-term test periods, 3–16 lead seizures during 169–364 days test period, whereas earlier studies did not differentiate between lead vs. non-lead seizures and used much shorter test periods (~few days long).
ARTICLE | doi:10.20944/preprints201611.0017.v1
Subject: Mathematics & Computer Science, Artificial Intelligence & Robotics Keywords: machine learning; unbalanced data; predicting rare events; NFL football; sports analytics
Online: 2 November 2016 (07:07:06 CET)
Turnovers in the National Football League (NFL) occur whenever a team loses possession of the ball due to a fumble, or an interception. Turnovers disrupt momentum of the offensive team, and represent lost opportunities to advance downfield and score. Teams with a positive differential turnover margin in a given game win $70\%$ of the time. Turnovers are statistically rare events, occurring apparently randomly. These characteristics make them difficult to predict. This investigation advances the hypothesis that turnovers are predictable in NFL football. Machine learning models are developed to learn the concept: At any point within a football game, what is the likelihood that a turnover will be observed on the next play from scrimmage? Results presented suggest evidence to support the working hypothesis. Under certain conditions, both fumbles and interceptions can be anticipated at low false discovery rates (less than $15\%$). This approach may be useful to inform in-game strategies to mitigate the negative consequences of turnovers by an offensive team, or to maximize their probability by a defensive squad.
ARTICLE | doi:10.20944/preprints201805.0283.v1
Subject: Engineering, Electrical & Electronic Engineering Keywords: venturini method; matrix converter; unbalanced voltage conditions; carrier-based pulse width modulation (PWM)
Online: 22 May 2018 (05:05:38 CEST)
Based on Venturini method, it is in favor of the modulation technique for controlling the matrix converter due to only use of the comparison between the duty cycles in time domain and the triangular carrier wave for generating the gating signals and the achievable voltage ratio between fundamental output magnitude and fundamental input magnitude to 0.866. However, even with simple modulation method and achieving maximum fundamental output magnitude, the possible input voltage unbalance conditions accordingly influence on the output performances (more reduction and distortion). Thus, a modified Venturini modulation method is presented in this paper, in order to solve the problems of unbalanced input voltage conditions on the matrix converter performances. The proposed strategy is to satisfy the desirable feature of the duty cycle modulating waves, as generated in the event of normal situation. Up to this approach, it can support either single-phase condition or two-phase condition. Performance of the proposed control strategy was verified by the simulated implementation in the MATLAB/Simulink software with showing good steady-state and dynamic operations.
ARTICLE | doi:10.20944/preprints201705.0136.v1
Subject: Engineering, Electrical & Electronic Engineering Keywords: generalized integrator; grid connected inverters; phase locked loops; renewable energy; symmetrical components; unbalanced voltage
Online: 18 May 2017 (04:40:16 CEST)
Frequency, amplitude and phase information of the grid voltage are of great importance in constructing a robust controller structure for grid connected inverter systems. This paper presents a simple and robust approach for the instantaneous estimation of positive and negative sequence voltage components under distorted voltage conditions. A second order generalized integrator (SOGI) is used to filter the distorted voltage and to generate orthogonal voltage components for each of the three phases. These filtered and orthogonal components are used for instantaneous calculation of symmetrical components. The implemented method is frequency adaptive; the method is demonstrated and compared to a conventional phase locked loop (PLL) technique with both MATLAB/Simulink simulations and experiments utilizing the dSPACE ds1103 digital controller.
ARTICLE | doi:10.20944/preprints202109.0181.v1
Subject: Keywords: Classification; stacking ensemble method; heart surgery; unbalanced data problem; hybrid predictive model; machine learning in healthcare; resampling method; Edited-Nearest-Neighbor; nonparametric test.
Online: 10 September 2021 (10:53:35 CEST)
Nowadays, according to spectacular improvement in health care and biomedical level, a tremendous amount of data is recorded by hospitals. In addition, the most effective approach to reduce disease mortality is to diagnose it as soon as possible. As a result, data mining by applying machine learning in the field of diseases provides good opportunities to examine the hidden patterns of this collection. An exact forecast of the mortality after heart surgery will cause Successful medical treatment and fewer costs. This research wants to recommend a new stacking predictive model after utilizing the random forest feature importance method to foresee the mortality after heart surgery on a highly unbalanced dataset by using the most practical features. To solve the unbalanced data problem, a combination of the SVM-SMOTE over-sampling algorithm and the Edited-Nearest-Neighbor under-sampling algorithm is used. This research compares the introduced model with some other machine learning classifiers to ensure efficiency through shuffle hold-out and 10-fold cross-validation strategies. In order to validate the performance of the implemented machine learning methods in this research, both shuffle hold-out, and 10-fold cross-validation results indicated that our model had the highest efficiency compared to the other models. Furthermore, the Friedman statistical test is applied to survey the differences between models. The result demonstrates that the introduced stacking model reached the most accurate predicting performance after Logistic Regression.