ABSTRACT Medical institutions are increasingly leveraging cloud servers to store electronic health records (EHRs), highlighting the need for robust data security measures to protect the sensitive personal information they contain. Our study introduces a blockchain‐enabled, fine‐grained data integrity auditing scheme that not only safeguards the confidentiality and integrity of EHRs within cloud‐based healthcare environments but also demonstrates a significant enhancement in data security with our statistical results, reinforcing the trustworthiness of cloud storage for sensitive medical data. The proposed scheme is notable for its support of dynamic user revocation, implementing a multi‐tiered role hierarchy that facilitates the efficient access revocation. In this hierarchy, adding new users or updating existing ones involves merely altering the edge labels, thereby obviating the need for a comprehensive recalculation of cryptographic keys. We have developed a smart contract‐based access control mechanism to ensure privacy while enabling granular access control. This mechanism leverages password and role‐based authentication to empower multi‐tiered roles with the ability to perform data integrity audits by their designated permissions. Through security analysis, we have substantiated that our protocol withstands attacks aimed at subversion, counterfeiting, and tag inconsistency. Compared to existing works, our approach uniquely integrates multi‐level role hierarchies with blockchain‐based dynamic revocation, achieving higher granularity and adaptability.

In modern microservices systems, identifying the root causes of faults in multivariate datasets is crucial.However, traditional methods that rely on statistical counts to detect anomalies can lead to excessive prediction errors due to too many attribute combinations without effectively utilizing the distribution characteristics of outliers.To address this, this paper introduce an improved multivariate root cause localization algorithm called JSqueeze.JSqueeze integrates Core Distance Bound Clustering method, a novel search algorithm, and an improved stopping strategy and introduces Jensen-Shannon divergence to assess the distribution properties of root causes.Based on this, a new two-phase algorithm framework is proposed.In the clustering phase, a new density clustering method is introduced, and in the root cause identification phase, a new robust heuristic search method is used.Experiments on a dataset with 6300 faults show that JSqueeze significantly improves the F1-score and time efficiency.Compared to existing state-of-the-art algorithms, our improved method increases the average F1-score by 5.8% and saves time by 14.5%.Moreover, JSqueeze shows superior performance in more complex datasets (i.e., those with multiple root causes), providing a new, robust, and efficient solution for anomaly localization in complex microservice systems.Code is available at https://github.com/DYamIK/JSqueeze.

In recent years, the convergence of driver monitoring systems (DMS), cloud technologies, and self-driving cars has gained increasing attention. Driver monitoring systems use a variety of camera and sensor technologies to detect the driver's state in real time, including fatigue levels, and stress levels based on emotional state. Such systems help to improve driving safety and reduce accidents caused by driver fatigue or emotional high pressure. This paper proposes an in-vehicle driver emotion recognition cloud computing scheme. Contact sensors are used in conjunction with a steering wheel to collect the electrical skin signals from the driver's palm and transmit the data to a cloud server via a 6G IoT device. The driver's emotional state such as fatigue, alertness, and stress level is measured by One-Class Support Vector Machines (OCSVM) combined with a Long Short-Term Memory (LSTM) recurrent neural network. When low mood, drowsiness, or lack of alertness are recognized, the driver is reminded and given feedback on safe driving through methods such as seat vibration and voice prompts. And the vehicle's automatic driving assistance mode will be turned on in extreme situations. The proposed method combines unsupervised learning with threshold-based wavelet denoising, effectively removing noise data generated by motion during the collection of human electrodermal signals. Experimental results demonstrate that the motion artifact removal algorithm presented in this paper exhibits superior denoising effects. Compared to traditional filtering algorithms, the SNR(Signal to Noise Ratio) is enhanced by 4.819 dB, while the RMSE(Root Mean Square Error) is reduced by 0.0385. Ultimately, the accuracy is improved by 2.44% compared to conventional emotion recognition methods.

Recently, phonetic posteriorgrams (PPGs) based methods have been quite popular in non-parallel singing voice conversion systems. However, due to the lack of acoustic information in PPGs, style and naturalness of the converted singing voices are still limited. To solve these problems, in this paper, we utilize an acoustic reference encoder to implicitly model singing characteristics. We experiment with different auxiliary features, including mel spectrograms, HuBERT, and the middle hidden feature (PPG-Mid) of pretrained automatic speech recognition (ASR) model, as the input of the reference encoder, and finally find the HuBERT feature is the best choice. In addition, we use contrastive predictive coding (CPC) module to further smooth the voices by predicting future observations in latent space. Experiments show that, compared with the baseline models, our proposed model can significantly improve the naturalness of converted singing voices and the similarity with the target singer. Moreover, our proposed model can also make the speakers with just speech data sing.

Singing voice conversion (SVC) aims to convert the voice of one singer to that of other singers while keeping the singing content and melody. On top of recent voice conversion works, we propose a novel model to steadily convert songs while keeping their naturalness and intonation. We build an end-to-end architecture, taking phonetic posteriorgrams (PPGs) as inputs and generating mel spectrograms. Specifically, we implement two separate encoders: one encodes PPGs as content, and the other compresses mel spectrograms to supply acoustic and musical information. To improve the performance on timbre and melody, an adversarial singer confusion module and a mel-regressive representation learning module are designed for the model. Objective and subjective experiments are conducted on our private Chinese singing corpus. Comparing with the baselines, our methods can significantly improve the conversion performance in terms of naturalness, melody, and voice similarity. Moreover, our PPG-based method is proved to be robust for noisy sources.

Automatic music transcription (AMT) is the task of transcribing audio recordings into symbolic representations. Recently, neural network based methods have been applied to AMT, and have achieved state-of-the-art results. However, many previous systems only detect onset and offset of notes in frame-wise, so the transcription resolution is limited to the frame hop size. There is a lack of research of using different strategies to encode onset and offset targets for training. In addition, previous AMT systems are sensitive to the misaligned onset and offset labels of audio recordings. Furthermore, there are limited research of sustain pedal transcription on large-scale datasets. In this article, we propose a high-resolution AMT system trained by regressing precise onset and offset times of piano notes. At inference, we propose an algorithm to analytically calculate the precise onset and offset times of piano notes and pedal events. We show that our AMT system is robust to misaligned onset and offset labels compared to previous systems. Our proposed system achieves an onset F1 of 96.72% on the MAESTRO dataset, outperforming previous onsets and frames system of 94.80%. Our system achieves a pedal onset F1 score of 91.86%, which is the first benchmark result on the MAESTRO dataset. We have released the source code and checkpoints of our work at https://github.com/bytedance/piano_transcription.

Multi-view learning has become a significant research topic in image processing, data mining and machine learning due to the proliferation of multi-view data. Considering the difficulty in obtaining labeled data in many real applications, we focus on the multi-view unsupervised feature selection problem. Most existing multi-view feature selection introduce an identical similarity matrix among different views, which cannot preserve the specific correlation between each single view. Also, some of these methods just consider either global or local structures. In this paper, we propose an embedding method, Adaptive Similarity Embedding for Unsupervised Multi-View Feature Selection (ASE-UMFS). This method reduces the high-dimensional data to the low dimensions and unifies different views to a combination weight matrix. We also use parameters to constraint the similarity matrix for the local structure, where the regularization term is used to add a prior of uniform distribution; taking into account of the independence in projection matrix among different views, optimization of the similarity matrix is further improved. To confirm the effectiveness of ASE-UMFS, comparisons are made with benchmark algorithm on real-world data sets. The experimental results demonstrate that the proposed algorithm outperforms several state-of-the-art methods in multi-view learning.