IFLBC: On the Edge Intelligence Using Federated Learning Blockchain Network

Lately there has been an increase in the number of Machine Learning (ML) and Artificial Intelligence (AI) applications ranging from recommendation systems to face to speech recognition. At the helm of the advent of deep learning is the proliferation of data from diverse data sources ranging from Internet-of-Things (IoT) devices to self-driving automobiles. Tapping into this unlimited reservoir of information presents the problem of finding quality data out of a myriad of irrelevant ones, which to this day, has been a significant issue in data science with a direct ramification of this being the inability to generate quality ML models for useful predictive analysis. Edge computing has been deemed a solution to some of issues such as privacy, security, data silos and latency, as it ventures to bring cloud computing services closer to end-nodes. A new form of edge computing known as edge-AI attempts to bring ML, AI, and predictive analytics services closer to the data source (end devices). In this paper, we investigate an approach to bring edge-AI to end-nodes through a shared machine learning model powered by the blockchain technology and a federated learning framework called iFLBC edge. Our approach addresses the issue of the scarcity of relevant data by devising a mechanism known as the Proof of Common Interest (PoCI) to sieve out relevant data from irrelevant ones. The relevant data is trained on a model, which is then aggregated along with other models to generate a shared model that is stored on the blockchain. The aggregated model is downloaded by members of the network which they can utilize for the provision of edge intelligence to end-users. This way, AI can be more ubiquitous as members of the iFLBC network can provide intelligence services to end-users.

[1]  Danda B. Rawat,et al.  Blockchain: Emerging Applications and Use Cases , 2019, ArXiv.

[2]  Peter Richtárik,et al.  Federated Optimization: Distributed Machine Learning for On-Device Intelligence , 2016, ArXiv.

[3]  Paramvir Bahl,et al.  Low Latency Geo-distributed Data Analytics , 2015, SIGCOMM.

[4]  Peter Richtárik,et al.  Federated Learning: Strategies for Improving Communication Efficiency , 2016, ArXiv.

[5]  Brent Waters,et al.  Fuzzy Identity-Based Encryption , 2005, EUROCRYPT.

[6]  Xiaofei Wang,et al.  Convergence of Edge Computing and Deep Learning: A Comprehensive Survey , 2019, IEEE Communications Surveys & Tutorials.

[7]  Michael I. Jordan,et al.  Latent Dirichlet Allocation , 2001, J. Mach. Learn. Res..

[8]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[9]  S. Fawcett,et al.  Data Science, Predictive Analytics, and Big Data: A Revolution that Will Transform Supply Chain Design and Management , 2013 .

[10]  Xu Chen,et al.  In-Edge AI: Intelligentizing Mobile Edge Computing, Caching and Communication by Federated Learning , 2018, IEEE Network.

[11]  Danda B. Rawat,et al.  Towards Federated Learning Approach to Determine Data Relevance in Big Data , 2019, 2019 IEEE 20th International Conference on Information Reuse and Integration for Data Science (IRI).

[12]  Seong-Lyun Kim,et al.  Blockchained On-Device Federated Learning , 2018, IEEE Communications Letters.

[13]  Weisong Shi,et al.  Edge Computing: Vision and Challenges , 2016, IEEE Internet of Things Journal.

[14]  Sokol Kosta,et al.  To offload or not to offload? The bandwidth and energy costs of mobile cloud computing , 2013, 2013 Proceedings IEEE INFOCOM.

[15]  Mariana Raykova,et al.  RapidChain: A Fast Blockchain Protocol via Full Sharding , 2018, IACR Cryptol. ePrint Arch..

[16]  Prateek Saxena,et al.  A Secure Sharding Protocol For Open Blockchains , 2016, CCS.

[17]  Christopher Frost,et al.  Spanner: Google's Globally-Distributed Database , 2012, OSDI.