scholarly journals Multi-Layer Random Perturbation Training for improving Model Generalization Efficiently

2021 ◽  
Author(s):  
Lis Kanashiro Pereira ◽  
Yuki Taya ◽  
Ichiro Kobayashi
Keyword(s):  
Random Perturbation ◽  
Perturbation Training ◽  
Model Generalization

scholarly journals Multi-Path and Group-Loss-Based Network for Speech Emotion Recognition in Multi-Domain Datasets

Sensors ◽  
2021 ◽  
Vol 21 (5) ◽  
pp. 1579 ◽  
Author(s):  
Kyoung Ju Noh ◽  
Chi Yoon Jeong ◽  
Jiyoun Lim ◽  
Seungeun Chung ◽  
Gague Kim ◽  
...  
Keyword(s):  
Emotion Recognition ◽  
Short Term Memory ◽  
Domain Adaptation ◽  
Classification Model ◽  
Speech Emotion Recognition ◽  
Target Domain ◽  
Model Generalization ◽  
Speech Database ◽  
Emotion Labels ◽  
Temporal Feature

Speech emotion recognition (SER) is a natural method of recognizing individual emotions in everyday life. To distribute SER models to real-world applications, some key challenges must be overcome, such as the lack of datasets tagged with emotion labels and the weak generalization of the SER model for an unseen target domain. This study proposes a multi-path and group-loss-based network (MPGLN) for SER to support multi-domain adaptation. The proposed model includes a bidirectional long short-term memory-based temporal feature generator and a transferred feature extractor from the pre-trained VGG-like audio classification model (VGGish), and it learns simultaneously based on multiple losses according to the association of emotion labels in the discrete and dimensional models. For the evaluation of the MPGLN SER as applied to multi-cultural domain datasets, the Korean Emotional Speech Database (KESD), including KESDy18 and KESDy19, is constructed, and the English-speaking Interactive Emotional Dyadic Motion Capture database (IEMOCAP) is used. The evaluation of multi-domain adaptation and domain generalization showed 3.7% and 3.5% improvements, respectively, of the F1 score when comparing the performance of MPGLN SER with a baseline SER model that uses a temporal feature generator. We show that the MPGLN SER efficiently supports multi-domain adaptation and reinforces model generalization.

scholarly journals A Novel Domain Adaptation-Based Intelligent Fault Diagnosis Model to Handle Sample Class Imbalanced Problem

Sensors ◽  
2021 ◽  
Vol 21 (10) ◽  
pp. 3382
Author(s):  
Zhongwei Zhang ◽  
Mingyu Shao ◽  
Liping Wang ◽  
Sujuan Shao ◽  
Chicheng Ma
Keyword(s):  
Fault Diagnosis ◽  
Domain Adaptation ◽  
Class Imbalance ◽  
Rolling Bearing ◽  
Industrial Applications ◽  
Mechanical Equipment ◽  
Reliable Operation ◽  
Cross Domain ◽  
Geometric Difference ◽  
Model Generalization

As the key component to transmit power and torque, the fault diagnosis of rotating machinery is crucial to guarantee the reliable operation of mechanical equipment. Regrettably, sample class imbalance is a common phenomenon in industrial applications, which causes large cross-domain distribution discrepancies for domain adaptation (DA) and results in performance degradation for most of the existing mechanical fault diagnosis approaches. To address this issue, a novel DA approach that simultaneously reduces the cross-domain distribution difference and the geometric difference is proposed, which is defined as MRMI. This work contains three parts to improve the sample class imbalance issue: (1) A novel distance metric method (MVD) is proposed and applied to improve the performance of marginal distribution adaptation. (2) Manifold regularization is combined with instance reweighting to simultaneously explore the intrinsic manifold structure and remove irrelevant source-domain samples adaptively. (3) The ℓ2-norm regularization is applied as the data preprocessing tool to improve the model generalization performance. The gear and rolling bearing datasets with class imbalanced samples are applied to validate the reliability of MRMI. According to the fault diagnosis results, MRMI can significantly outperform competitive approaches under the condition of sample class imbalance.

scholarly journals A bat optimization algorithm with moderate orientation and perturbation of trend

2021 ◽  
Vol 15 ◽  
pp. 174830262110084
Author(s):  
Jingsen Liu ◽  
Hongyuan Ji ◽  
Qingqing Liu ◽  
Yu Li
Keyword(s):  
Local Search ◽  
Optimization Algorithm ◽  
Random Perturbation ◽  
Bat Algorithm ◽  
Local Extremum ◽  
Global Search ◽  
Deep Mining ◽  
Nonlinear Variation ◽  
Optimal Value ◽  
Search Stage

In order to improve the convergence speed and optimization accuracy of the bat algorithm, a bat optimization algorithm with moderate optimal orientation and random perturbation of trend is proposed. The algorithm introduces the nonlinear variation factor into the velocity update formula of the global search stage to maintain a high diversity of bat populations, thereby enhanced the global exploration ability of the algorithm. At the same time, in the local search stage, the position update equation is changed, and a strategy that towards optimal value modestly is used to improve the ability of the algorithm to local search for deep mining. Finally, the adaptive decreasing random perturbation is performed on each bat individual that have been updated in position at each generation, which can improve the ability of the algorithm to jump out of the local extremum, and to balance the early global search extensiveness and the later local search accuracy. The simulating results show that the improved algorithm has a faster optimization speed and higher optimization accuracy.

scholarly journals Random perturbation of low rank matrices: Improving classical bounds

2018 ◽  
Vol 540 ◽  
pp. 26-59 ◽  
Author(s):  
Sean O'Rourke ◽  
Van Vu ◽  
Ke Wang
Keyword(s):  
Random Perturbation ◽  
Low Rank

Velocity log upscaling based on reversible jump Markov chain Monte Carlo simulated annealing

Geophysics ◽  
2016 ◽  
Vol 81 (5) ◽  
pp. R293-R305 ◽  
Author(s):  
Sireesh Dadi ◽  
Richard Gibson ◽  
Kainan Wang
Keyword(s):  
Monte Carlo ◽  
Markov Chain ◽  
Simulated Annealing ◽  
Markov Chain Monte Carlo ◽  
Random Perturbation ◽  
Current Model ◽  
Sampling Technique ◽  
Maximum Deviation ◽  
Reversible Jump ◽  
Layer Boundary

Upscaling log measurements acquired at high frequencies and correlating them with corresponding low-frequency values from surface seismic and vertical seismic profile data is a challenging task. We have applied a sampling technique called the reversible jump Markov chain Monte Carlo (RJMCMC) method to this problem. A key property of our approach is that it treats the number of unknowns itself as a parameter to be determined. Specifically, we have considered upscaling as an inverse problem in which we considered the number of coarse layers, layer boundary depths, and material properties as the unknowns. The method applies Bayesian inversion, with RJMCMC sampling and uses simulated annealing to guide the optimization. At each iteration, the algorithm will randomly move a boundary in the current model, add a new boundary, or delete an existing boundary. In each case, a random perturbation is applied to Backus-average values. We have developed examples showing that the mismatch between seismograms computed from the upscaled model and log velocities improves by 89% compared to the case in which the algorithm is allowed to move boundaries only. The layer boundary distributions after running the RJMCMC algorithm can represent sharp and gradual changes in lithology. The maximum deviation of upscaled velocities from Backus-average values is less than 10% with most of the values close to zero.

scholarly journals Second-order differential equations with random perturbations and small parameters

2017 ◽  
Vol 147 (4) ◽  
pp. 763-779
Author(s):  
M. Kamenskii ◽  
S. Pergamenchtchikov ◽  
M. Quincampoix
Keyword(s):  
Differential Equations ◽  
Small Parameter ◽  
Random Perturbation ◽  
Stochastic Averaging ◽  
Second Order ◽  
Random Perturbations ◽  
Existence And Uniqueness Theorem ◽  
Second Order Differential Equations ◽  
Averaging Theorem ◽  
Small Parameters

We consider boundary-value problems for differential equations of second order containing a Brownian motion (random perturbation) and a small parameter and prove a special existence and uniqueness theorem for random solutions. We study the asymptotic behaviour of these solutions as the small parameter goes to zero and show the stochastic averaging theorem for such equations. We find the explicit limits for the solutions as the small parameter goes to zero.

scholarly journals Can Treadmill Slip-Perturbation Training Reduce Longer-Term Fall Risk Upon Overground Slip Exposure?

2020 ◽  
Vol 36 (5) ◽  
pp. 298-306 ◽  
Author(s):  
Anna Lee ◽  
Tanvi Bhatt ◽  
Xuan Liu ◽  
Yiru Wang ◽  
Shuaijie Wang ◽  
...  
Keyword(s):  
Older Adults ◽  
Fall Risk ◽  
Center Of Mass ◽  
Control Group ◽  
Initial Training ◽  
Single Session ◽  
Control Groups ◽  
Perturbation Training ◽  
Positive Effect ◽  
Better Than

The purpose was to examine and compare the longer-term generalization between 2 different practice dosages for a single-session treadmill slip-perturbation training when reexposed to an overground slip 6 months later. A total of 45 older adults were conveniently assigned to either 24 or 40 slip-like treadmill perturbation trials or a third control group. Overground slips were given immediately after initial training, and at 6 months after initial training in order to examine immediate and longer-term effects. The performance (center of mass stability and vertical limb support) and fall percentage from the laboratory-induced overground slips (at initial posttraining and at 6 mo) were measured and compared between groups. Both treadmill slip-perturbation groups showed immediate generalization at the initial posttraining test and longer-term generalization at the 6-month retest. The higher-practice-dosage group performed significantly better than the control group (P < .05), with no difference between the lower-practice-dosage and the control groups at the 6-month retest (P > .05). A single session of treadmill slip-perturbation training showed a positive effect for reducing older adults’ fall risk for laboratory-induced overground slips. A higher-practice dosage of treadmill slip perturbations could be more beneficial for further reducing fall risk.

scholarly journals EFFICACY OF ROLLING BOARD PERTURBATION TRAINING ON BALANCE AMONG HEMIPARETIC STROKE PATIENTS - A COMPARATIVE STUDY

2020 ◽  
Vol 6 (1) ◽  
Author(s):  
Dr. Senthil kumar ◽  
Dr. Franklin Shaju M.K m k ◽  
Dr. Vijaya Senthil Kumar kumar ◽  
Dr. A. velmurugan
Keyword(s):  
Public Health Problem ◽  
Berg Balance Scale ◽  
Major Public Health Problem ◽  
Stroke Patients ◽  
Perturbation Training ◽  
Before And After ◽  
Group A ◽  
Hemiparetic Stroke ◽  
Conventional Physiotherapy ◽  
Group B

Background of the study: Stroke is a major public health problem that ranks in the top four causes of death in most of the countries and is responsible for a large proportion of the burden of neurologic disorders. Patients with stroke have poor balance because they cannot control dynamically the size of the base of support or the location of the line of gravity. Perturbation training undergoes the maximal sway possible without losing his balance. Objective of the study: The objective of the study is to find the effects of rolling board perturbation training on balance among hemiparetic stroke patients. Methodology: Thirty clinically diagnosed hemiparetic stroke patients were selected based on the inclusion and exclusion criteria. They were randomly allocated into two groups (Group A and Group B) consists of 15 subjects each. Group A received conventional physiotherapy alone and group B received rolling board perturbation training along with conventional physiotherapy. Intervention lasted for 4 weeks, three days in a week and one hour per day. Balance was measured before and after 4 weeks of intervention by berg balance scale. Conclusion: Both conventional physiotherapy alone and rolling board perturbation training along with conventional physiotherapy significantly improved balance among hemiparetic stroke patients. When comparing both rolling board perturbation training along with conventional physiotherapy is more effective than conventional physiotherapy alone in improving balance among hemiparetic stroke patients.

scholarly journals Generalized Prediction of Shock in Intensive Care Units using Deep Learning

2021 ◽  
Author(s):  
Aditya Nagori ◽  
Anushtha Kalia ◽  
Arjun Sharma ◽  
Pradeep Singh ◽  
Harsh Bandhey ◽  
...  
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Prediction Models ◽  
Age Groups ◽  
Receiver Operating Curve ◽  
Learning Approach ◽  
New Delhi ◽  
Model Generalization ◽  
Pediatric Icu ◽  
Geographical Context

Shock is a major killer in the ICU and machine learning based early predictions can potentially save lives. Generalization across age and geographical context is an unaddressed challenge. In this retrospective observational study, we built real-time shock prediction models generalized across age groups and continents. More than 1.5 million patient-hours of novel data from a pediatric ICU in New Delhi and 5 million patient-hours from the adult ICU MIMIC database were used to build models. We achieved model generalization through a novel fractal deep-learning approach and predicted shock up to 12 hours in advance. Our deep learning models showed a receiver operating curve (AUROC) drop from 78% (95%CI, 73-83) on MIMIC data to 66% (95%CI, 54-78) on New Delhi data, outperforming standard machine learning by nearly a 10% gap. Therefore, better representations and deep learning can partly address the generalizability-gap of ICU prediction models trained across geographies. Our data and algorithms are publicly available as a pre-configured docker environment at https://github.com/SAFE-ICU/ShoQPred.

Sign in / Sign up

Export Citation Format

Share Document