This article reviews how fully homomorphic encryption can be combined with federated learning to enhance privacy protection, discusses recent FHE advancements, presents practical system designs such as POSEIDON, and outlines open research challenges and future research directions.
The award‑winning solution combines a GPU‑accelerated TEE framework (HyperGPU) and a multi‑GPU zkSNARK acceleration scheme (DistMSM) to provide fast, privacy‑preserving AI inference proofs, earning the third‑place Innovation Team prize at the 2024 Financial Cipher Cup competition.
The Ant Shared Intelligence workshop at ACM CCS 2020 invites researchers and practitioners to submit short papers on privacy‑preserving machine learning techniques such as secure multi‑party computation, homomorphic encryption, differential privacy, federated learning, and related applications, with a submission deadline of June 21, 2020.
At the 32nd USENIX Security Symposium in Anaheim, Ant Group’s Research Institute sponsored the event and showcased two first‑author papers—one introducing the Squirrel framework for fast, secure two‑party computation of Gradient Boosting Decision Trees, and another proposing an efficient 3‑party protocol for binary circuits in maliciously‑secure DNN inference.
The article explains how Ant Group tackles the privacy‑usability trade‑off in AI by combining Trusted Execution Environments and Multi‑Party Computation into a “shared intelligence” framework, contrasting it with federated learning, detailing technical architectures, training workflows, and its impact on financial data sharing.
The article explains how Trusted Execution Environments (TEE), built on ARM TrustZone, provide a secure world separate from the Rich Execution Environment, detailing its architecture, GP API interactions, and how fingerprint enrollment and authentication are performed within TEE to protect sensitive biometric data.
This article explains how privacy computing technologies such as federated learning, multi‑party computation, and trusted execution environments, combined with blockchain, address data sharing challenges in the digital economy by protecting privacy, ensuring compliance, and enabling secure, trusted collaboration across enterprises and government agencies.
Federated learning enables multiple organizations to jointly train high‑performing AI models without sharing raw data, using techniques such as secure multi‑party computation, differential privacy, and homomorphic encryption, thereby overcoming data‑island and regulatory constraints and supporting applications in mobile edge AI, finance, retail, and healthcare.
Occlum v1.0, the open‑source trusted execution environment operating system released by Ant Group, delivers up to five‑fold performance improvements, supports over 150 Linux syscalls, introduces async I/O, dynamic memory management, and a Spark‑BigDL big‑data analysis solution, while outlining future GPU and TDX extensions.
This presentation reviews recent research on data security and utilization in the large‑model era, covering privacy‑preserving federated learning, knowledge‑transfer techniques, prototype‑based modeling, multi‑model fusion methods such as FuseGen, and introduces the federated knowledge computing platform FLAIR for both horizontal and vertical federated scenarios.
This article introduces the background of privacy computing, explains the three‑stage FIRM reference architecture for federated learning, describes key technologies such as the Ionic Bond communication framework and HeteroDeepFM, and showcases real‑world applications in marketing, risk control, and government sectors.
Privacy-Preserving Machine Learning (PPML) integrates cryptographic techniques such as federated learning, differential privacy, homomorphic encryption, and secure multi-party computation to enable model training and inference on encrypted or distributed data, thereby breaking data silos while safeguarding privacy across sectors like healthcare, finance, and advertising.
This article presents sixteen peer‑reviewed research papers published in top conferences and journals in 2024, covering trusted AI, large‑model applications, network security, adversarial training, deep‑fake detection, secure inference, and related topics from collaborations among Alibaba, Tsinghua, Zhejiang, and other leading institutions.
This article reviews the evolution of federated learning, outlines its algorithms and standards, proposes centralized and decentralized network‑intelligence architectures for mobile communications, and presents a customer‑experience‑management case study that demonstrates how federated learning improves model accuracy and privacy across multiple regional nodes.
This article explains how data islands hinder AI progress, introduces federated learning as a privacy‑preserving solution, details JD.com’s 9N‑FL platform—including its architecture, features, classification, privacy‑preserving techniques, and algorithm support—and demonstrates its successful application in e‑commerce advertising that yielded over 15% revenue growth.
The article compares shared intelligence and federated learning, examines privacy‑preserving techniques such as MPC, TEE, and differential privacy, discusses gradient‑inversion attacks and their mitigations, and presents Ant Group’s end‑to‑end system design and real‑world deployments in finance.
