Novel Time Domain Based Upper-Limb Prosthesis Control using Incremental Learning Approach

Pancholi, Sidharth; Joshi, Amit M.

doi:10.48550/arxiv.2109.04194

Cited by 2 publications

(2 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In some recent years, researchers have explored the potential of several deep learning and machine learning methods for brain machine interface applications [57]- [60]. They have been used extensively for the evaluation of mental illnesses.…”

Section: Discussionmentioning

confidence: 99%

Novel EEG based Schizophrenia Detection with IoMT Framework for Smart Healthcare

Sharma¹,

Joshi²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

In the field of neuroscience, Brain activity analysis is always considered as an important area. Schizophrenia(Sz) is a brain disorder that severely affects the thinking, behavior, and feelings of people all around the world. Electroencephalography (EEG) is proved to be an efficient biomarker in Sz detection. EEG is a non-linear time-series signal and utilizing it for investigation is rather crucial due to its non-linear structure. This paper aims to improve the performance of EEG based Sz detection using a deep learning approach. A novel hybrid deep learning model known as SzHNN (Schizophrenia Hybrid Neural Network), a combination of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) has been proposed. CNN network is used for local feature extraction and LSTM has been utilized for classification. The proposed model has been compared with CNN only, LSTM only, and machine learning based models. All the models have been evaluated on two different datasets wherein Dataset 1 consists of 19 subjects and Dataset 2 consists of 16 subjects. Several experiments have been conducted for the same using various parametric settings on different frequency bands and using different sets of electrodes on the scalp. Based on all the experiments, it is evident that the proposed hybrid model (SzHNN) provides the highest classification accuracy of 99.9% in comparison to other existing models. The proposed model overcomes the influence of different frequency bands and even showed a much better accuracy of 91% with only 5 electrodes. The proposed model is also evaluated in Internet of Medical Things (IoMT) framework for smart healthcare and remote monitoring applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Novel EEG based Schizophrenia Detection with IoMT Framework for Smart Healthcare

Sharma¹,

Joshi²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…For example, Amrollahi et al proposed a continual learning algorithm that can incrementally learn from new patient data for sepsis prediction [21]. Continual learning has also been studied in other fields, such as navigation [22], text-to-speech synthesis [23] and prosthetic control [24], to just name a few. In the field of navigation for instance, a robot needs to adapt when facing different navigation environments, a problem known as lifelong navigation [22].…”

Section: Related Workmentioning

confidence: 99%

Lifelong Adaptive Machine Learning for Sensor-Based Human Activity Recognition Using Prototypical Networks

Thomaz

2022

Sensors

View full text Add to dashboard Cite

Continual learning (CL), also known as lifelong learning, is an emerging research topic that has been attracting increasing interest in the field of machine learning. With human activity recognition (HAR) playing a key role in enabling numerous real-world applications, an essential step towards the long-term deployment of such systems is to extend the activity model to dynamically adapt to changes in people’s everyday behavior. Current research in CL applied to the HAR domain is still under-explored with researchers exploring existing methods developed for computer vision in HAR. Moreover, analysis has so far focused on task-incremental or class-incremental learning paradigms where task boundaries are known. This impedes the applicability of such methods for real-world systems. To push this field forward, we build on recent advances in the area of continual learning and design a lifelong adaptive learning framework using Prototypical Networks, LAPNet-HAR, that processes sensor-based data streams in a task-free data-incremental fashion and mitigates catastrophic forgetting using experience replay and continual prototype adaptation. Online learning is further facilitated using contrastive loss to enforce inter-class separation. LAPNet-HAR is evaluated on five publicly available activity datasets in terms of its ability to acquire new information while preserving previous knowledge. Our extensive empirical results demonstrate the effectiveness of LAPNet-HAR in task-free CL and uncover useful insights for future challenges.

show abstract

Novel Time Domain Based Upper-Limb Prosthesis Control using Incremental Learning Approach

Cited by 2 publications

References 37 publications

Novel EEG based Schizophrenia Detection with IoMT Framework for Smart Healthcare

Novel EEG based Schizophrenia Detection with IoMT Framework for Smart Healthcare

Lifelong Adaptive Machine Learning for Sensor-Based Human Activity Recognition Using Prototypical Networks

Contact Info

Product

Resources

About