Deep learning approaches are generally complex, requiring extensive computational resources and having high time complexity. Transfer learning is a state-of-the-art approach to reducing the requirements of high computational resources by using pre-trained models without compromising accuracy and performance. In conventional studies, pre-trained models are trained on datasets from different but similar domains with many domain-specific features. The computational requirements of transfer learning are directly dependent on the number of features that include the domain-specific and the generic features. This article investigates the prospects of reducing the computational requirements of the transfer learning models by discarding domain-specific features from a pre-trained model. The approach is applied to breast cancer detection using the dataset curated breast imaging subset of the digital database for screening mammography and various performance metrics such as precision, accuracy, recall, F1-score, and computational requirements. It is seen that discarding the domain-specific features to a specific limit provides significant performance improvements as well as minimizes the computational requirements in terms of training time (reduced by approx. 12%), processor utilization (reduced approx. 25%), and memory usage (reduced approx. 22%). The proposed transfer learning strategy increases accuracy (approx. 7%) and offloads computational complexity expeditiously.
Download full-text PDF |
Source |
---|---|
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC11041970 | PMC |
http://dx.doi.org/10.7717/peerj-cs.1938 | DOI Listing |
Chemistry
December 2024
Pandit Deendayal Energy University, Chemistry, Gandhinagar, Gujarat-382077, India, Gandhinagar, INDIA.
The accurate discrimination among various volatile organic compounds, especially ethanol and acetone possess a serious concern for metal oxide based chemiresistive sensors. The work presents a systematic approach to address the issue by utilizing superior sensing potentiality of Zn0.5Ni0.
View Article and Find Full Text PDFJAMA Netw Open
December 2024
Division of Geriatrics, Department of Medicine, University of California, San Francisco.
JAMA Netw Open
December 2024
Department of Behavioural Science and Health, Institute of Epidemiology and Health Care, University College London, London, United Kingdom.
Importance: Issues related to social connection are increasingly recognized as a global public health priority. However, there is a lack of a holistic understanding of social connection and its health impacts given that most empirical research focuses on a single or few individual concepts of social connection.
Objective: To explore patterns of social connection and their associations with health and well-being outcomes.
Int Urol Nephrol
December 2024
Department of Thoracic Surgery, West China Hospital, Sichuan University, No. 37, Guoxue Lane, Wuhou District, Chengdu, 610041, Sichuan Province, China.
This paper evaluated the bibliometric study by Li et al. (Int Urol Nephrol, 2024) on machine learning in renal medicine. Although the study claims to summarize the forefront trends and hotspots in this field, several key issues require further clarification to effectively guide future research.
View Article and Find Full Text PDFEmerg Radiol
December 2024
Emergency Radiology, Department of Radiology, Massachusetts General Hospial, Boston, USA.
Background: Emergency/trauma radiology artificial intelligence (AI) is maturing along all stages of technology readiness, with research and development (R&D) ranging from data curation and algorithm development to post-market monitoring and retraining.
Purpose: To develop an expert consensus document on best research practices and methodological priorities for emergency/trauma radiology AI.
Methods: A Delphi consensus exercise was conducted by the ASER AI/ML expert panel between 2022-2024.
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!