No abstract

Advancements in Industry 4.0 brought tremendous improvements in the healthcare sector, such as better quality of treatment, enhanced communication, remote monitoring, and reduced cost. Sharing healthcare data with healthcare providers is crucial for harnessing the benefits of such improvements. In general, healthcare data holds sensitive information about individuals. Hence, sharing such data is challenging because of various security and privacy issues. According to privacy regulations and ethical requirements, it is essential to preserve the privacy of patients before sharing data for medical research. State-of-the-art literature on privacy preserving studies either uses cryptographic approaches to protect the privacy or uses anonymizing techniques regardless of the type of attributes, this results in poor protection and data utility. In this paper, we propose an attribute-focused privacy preserving data publishing scheme. The proposed scheme is two-fold, comprising a fixed-interval approach to protect numerical attributes and an improved l-diverse slicing approach to protect the categorical and sensitive attributes. In the fixed-interval approach, the original values of the healthcare data are replaced with an equivalent computed value. The improved l-diverse slicing approach partitions the data both horizontally and vertically to avoid privacy leaks. Extensive experiments with real-world datasets are conducted to evaluate the performance of the proposed scheme. The classification models built on anonymized dataset yields approximately 13% better accuracy than benchmarked algorithms. Experimental analyses show that the average information loss which is measured by normalized certainty penalty (NCP) is reduced by 12% compared to similar approaches. The attribute focused scheme not only provides data utility but also prevents the data from membership disclosures, attribute disclosures, and identity disclosures.

Word-level sign language recognition (WSLR) is the backbone for continuous sign language recognition (CSLR) that infers glosses from sign videos. Finding the relevant gloss from the sign sequence and detecting explicit boundaries of the glosses from sign videos is a persistent challenge. In this paper, we propose a systematic approach for gloss prediction in WLSR using the Sign2Pose Gloss prediction transformer model. The primary goal of this work is to enhance WLSR’s gloss prediction accuracy with reduced time and computational overhead. The proposed approach uses hand-crafted features rather than automated feature extraction, which is computationally expensive and less accurate. A modified key frame extraction technique is proposed that uses histogram difference and Euclidean distance metrics to select and drop redundant frames. To enhance the model’s generalization ability, pose vector augmentation using perspective transformation along with joint angle rotation is performed. Further, for normalization, we employed YOLOv3 (You Only Look Once) to detect the signing space and track the hand gestures of the signers in the frames. The proposed model experiments on WLASL datasets achieved the top 1% recognition accuracy of 80.9% in WLASL100 and 64.21% in WLASL300. The performance of the proposed model surpasses state-of-the-art approaches. The integration of key frame extraction, augmentation, and pose estimation improved the performance of the proposed gloss prediction model by increasing the model’s precision in locating minor variations in their body posture. We observed that introducing YOLOv3 improved gloss prediction accuracy and helped prevent model overfitting. Overall, the proposed model showed 17% improved performance in the WLASL 100 dataset.

No abstract

No abstract