Multitentacle Federated Learning Over Software-Defined Industrial Internet of Things Against Adaptive Poisoning Attacks

Abstract

While Federated learning (FL) is attractive for pulling privacy-preserving distributed training data, the credibility of participating clients and non-inspectable data pose new security threats, of which poisoning attacks are particularly rampant and hard to defend without compromising privacy, performance or other desirable properties of FL. To tackle this problem, we propose a self-purified FL (SPFL) method that enables benign clients to exploit trusted historical features of locally purified model to supervise the training of aggregated model in each iteration. The purification is performed by an attention-guided self-knowledge distillation where the teacher and student models are optimized locally for task loss, distillation loss and attention-based loss simultaneously. SPFL imposes no restriction on the communication protocol and aggregator at the server. It can work in tandem with any existing secure aggregation algorithms and protocols for augmented security and privacy guarantee. We experimentally demonstrate that SPFL outperforms state-of-the-art FL defenses against various poisoning attacks. The attack success rate of SPFL trained model is at most 3% above that of a clean model, even if the poisoning attack is launched in every iteration with all but one malicious clients in the system. Meantime, it improves the model quality on normal inputs compared to FedAvg, either under attack or in the absence of an attack.

A Survey of Trusted Computing Solutions Using FPGAs

“…Finally, deep learning technology may challenge traditional trusted computing, where FPGAs are also involved on the application side. As future work, we consider exploring AI-based security technologies and how they match with the FPGA architecture [104].…”

Communications Security in Industry X: A Survey

“…For example, one of the main threats that machine learning can induce in the systems is the denial of detection (DoD). The DoD can prevent machine learning from generating signals, for instance, from events, failures, and even cyberattacks using adversarial examples [239] and data poisoning [240]. Another threat that machine learning can induce is leaking sensitive information from the company or factory.…”

Information-Centric Mobile Networks: A Survey, Discussion, and Future Research Directions

“…Moreover Since the NDN caching mechanism stores the consumer-requested data (comprised of both sensitive and nonsensitive information) at multiple static and mobile entities distributed across the network, securing such data is highly essential to avoid privacy leakage, poisoning, or backdoor attacks. Federated learning-backed frameworks such as multi-tentacle federated learning (MTFL) [117] can be adapted to reduce the chances of aforementioned attacks and ensure the trustiness of stored data on static and mobile nodes across various network nodes.…”