Fully Decentralized Semi-supervised Learning via Privacy-preserving Matrix Completion

Fierimonte, Roberto; Scardapane, Simone; Uncini, Aurelio; Panella, Massimo

doi:10.1109/tnnls.2016.2597444

Cited by 46 publications

(19 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, an independent cloud server outside the sites is necessary. Reference [35] proposed a distributed matrix completion algorithm and used it to perform a distributed semisupervised manifold regularization. During the matrix completion, each site exchanges inter-site data similarities with neighbors through privacy-preserving similarity computation protocols.…”

Section: Related Workmentioning

confidence: 99%

“…During the matrix completion, each site exchanges inter-site data similarities with neighbors through privacy-preserving similarity computation protocols. According to [36], a distance-recoverable protocol is not secure against malicious adversaries, which makes the method in [35] only suitable against a semi-honest adversary. Reference [37] assumed a scenario where additional nonprivate data are available and provided a learning model to improve the accuracy of a differential-private classifier using both private and non-private data.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Mixture-Model-Based Graph for Privacy-Preserving Semi-Supervised Learning

Yang

2020

IEEE Access

View full text Add to dashboard Cite

Privacy has become a major concern in data mining as it is utilized in many important applications. Distributed privacy-preserving data mining (DPPDM) is one of the techniques to address this concern, which focuses on protecting private information of members in distributed systems during data mining. As DPPDM is widely discussed in recent works, the semi-supervised manner of learning still draws less attention in this field. In this paper, a mixture-model-based semi-supervised DPPDM method is proposed. By introducing our method, a site in a distributed system is able to initiate a learning process using labeled data of its own and unlabeled data from all the sites. During the process, no individual data of any site is revealed to others, no information about data can be traced back to any specific site, and only the initiating site learns the result. We propose a parameter-masking privacy-preserving Expectation-Maximization (EM) algorithm and a mixture-model-based semi-supervised learning algorithm as the two main steps of our method. Experiments on both synthetic and real-world data demonstrate the effectiveness of the proposed method. INDEX TERMS Data privacy, distributed computing, expectation-maximization algorithms, graph theory, semisupervised learning. • The objective of the classifier. Different groups of categories are allowed to be introduced to label the same data

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Mixture-Model-Based Graph for Privacy-Preserving Semi-Supervised Learning

Yang

2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Therefore, as the LLDs cannot be reconstructed, the contents of the original speech signals are protected. Recently, a decentralized SSL paradigm was proposed in [137], in which privacy-preserving matrix completion algorithms are used, so that only learned knowledge is transferred between different clients, while the raw data are incommutable. However, as these approaches cannot fully guarantee client security and privacy or maintain the original performance, continued research addressing privacy concerns is required.…”

Section: Conclusion and Challenges For Future Workmentioning

confidence: 99%

Advanced Data Exploitation in Speech Analysis: An overview

Zhang

Cummins

Schuller

2017

IEEE Signal Process. Mag.

View full text Add to dashboard Cite

With recent advances in machine-learning techniques for automatic speech analysis (ASA)-the computerized extraction of information from speech signals-there is a greater need for high-quality, diverse, and very large amounts of data. Such data could be game-changing in terms of ASA system accuracy and robustness, enabling the extraction of feature representations or the learning of model parameters immune to confounding factors, such as acoustic variations, unrelated to the task at hand. However, many current ASA data sets do not meet the desired properties. Instead, they are often recorded under less than ideal conditions, with the corresponding labels sparse or unreliable.In addressing these issues, this article provides a comprehensive overview of state-of-the-art ASA data exploitation techniques that have been developed to take advantage of knowledge gained from related but unlabeled or different data sources to improve the performance of a particular ASA task of interest. We first identify three primary data challenges: sparse, unreliable, and unmatched data. We then review the corresponding approaches. The conditions, advantages, and drawbacks of using a range of differing data-mining techniques are also discussed. Finally, other data challenges and potential future research directions in this field are presented. OpportunitiesTraditionally, tasks such as data collection and annotation have been performed by small groups of experts in a laboratory setting. This conventional work paradigm is often tedious, time consuming, and costly. However, the ongoing information and communication technologies revolution and related technologies, such as the Internet of Things (IoT) and cloud computing, are providing us with opportunities to exploit larger amounts of speech data in more effective ways than ever before.The IoT, as a global infrastructure of the information society, is expected to offer advanced services (i.e., data collection) by interconnecting a wide variety of contemporary recording devices, such as smartphones, wearable devices, and tablets. Furthermore, as these devices often have microphones, social media apps, and Internet connectivity, they can be considered distributed sensors or entryways for speech collection and processing. Thus, the advance of Internet technologies and the ubiquity of smart devices can drastically reduce the cost and time associated with collecting and processing speech data.Cloud computing, or Internet-based computing, is expected to provide an on-demand computing resource. Thus, it gives an opportunity to store, access, and analyze the volume of speech data generated by the distributed devices mentioned previously. Cloud computing has been shown not only to minimize the costs associated with an ever-increasing demand for greater computational resources but also to reduce the cost associated with infrastructure maintenance and user access. Motivated by these advantages, most major speech technology providers have already shifted their primary research and application attentio...

show abstract

“…In this work, we are exploring the use of compression in a framework for distributed classification. The main motivation of this work is the optimization of the distributed machine learning algorithm when computational and communication costs are to be preserved as is the case in, e.g., resource-constrained scenarios (e.g., edge machine learning [43], smart sensing and privacy-preserving algorithms [8]). Broadly, the objective for compression can be formulated as…”

Section: Introductionmentioning

confidence: 99%

On Effects of Compression with Hyperdimensional Computing in Distributed Randomized Neural Networks

Rosato

Panella

Osipov

et al. 2021

Advances in Computational Intelligence

Self Cite

View full text Add to dashboard Cite

A change of the prevalent supervised learning techniques is foreseeable in the near future: from the complex, computational expensive algorithms to more flexible and elementary training ones. The strong revitalization of randomized algorithms can be framed in this prospect steering. We recently proposed a model for distributed classification based on randomized neural networks and hyperdimensional computing, which takes into account cost of information exchange between agents using compression. The use of compression is important as it addresses the issues related to the communication bottleneck, however, the original approach is rigid in the way the compression is used. Therefore, in this work, we propose a more flexible approach to compression and compare it to conventional compression algorithms, dimensionality reduction, and quantization techniques.

show abstract

Fully Decentralized Semi-supervised Learning via Privacy-preserving Matrix Completion

Cited by 46 publications

References 48 publications

Mixture-Model-Based Graph for Privacy-Preserving Semi-Supervised Learning

Mixture-Model-Based Graph for Privacy-Preserving Semi-Supervised Learning

Advanced Data Exploitation in Speech Analysis: An overview

On Effects of Compression with Hyperdimensional Computing in Distributed Randomized Neural Networks

Contact Info

Product

Resources

About