Sherpa.ai Federated Learning Platform
Sherpa.ai Privacy-Preserving Federated Learning Platform is a state of the art distributed platform for homogeneous and heterogeneous data model training providing the user with a complete control of its data, implementing a set of defenses tailored for distributed environments and with a comprehensive and seamless UI.
SHERPA.AI
PRIVACY-PRESERVING PLATFORM
The first cloud based Privacy-preserving B2B SaaS platform.
The 3 Federated Learning paradigms (Horizontal, Vertical and Transfer).
Integrated Privacy-Enhancing complementary technologies (Differential Privacy, Homomorphic Encryption, Secure Multiparty Computation, among others) that make our platform the most advanced and robust.
Pluggable framework to streamline deployment and with an intuitive and user-friendly UI that will help to democratize the access to data privacy to a wider spectrum.
UNLOCK NEW SCENARIOS WITH PRIVACY-PRESERVING AI
FEDERATED LEARNING PARADIGMS
HORIZONTAL FEDERATED LEARNING
Same type of data in different nodes
VERTICAL FEDERATED LEARNING
Different type of data in different nodes
FEDERATED TRANSFER LEARNING
The nodes share a few samples of information
ALGORITHMS AND
AI MODEL COVERAGE
FLEXIBILITY AND SCALABILITY.
Adaptability of Sherpa.ai solution to most ML models
The Sherpa.ai Federated Learning platform has been designed to be highly flexible, to adapt to a wide variety of AI algorithms and models. Our platform is ready for user customized models. Furthermore, the Sherpa.ai technology is scalable to a wide range of possibilities, corresponding to different fields of application. Examples of algorithms and ai model coverage are available below.
BUILD YOUR MODEL
IN ANY FRAMEWORK
Sherpa.ai platform fits your model.
Its the most flexible and open platform, fully framework agnostic, supporting among others: scikit-learn and TensorFlow. Any existent model can be fitted to Sherpa.ai FL platform.
TRAIN WITH ANY
DATA SOURCE
Any Python supported datasource can be used.
By implementing a set of interfaces the node can access the training data on its original sources. File access through local or remote filesystems, relational and no-SQL databases, HTTP data requests... Exploit your data wherever it is stored.
PLUGABLE FRAMEWORKS
(FAST DEPLOYMENT)
The agility of a platform can be estimated, among other things, by its ability to adapt to existing models. Thus, offering interfaces to existing frameworks is a major competitive advantage.
NODE INSTALLATION BASED ON DOCKER CONTAINERS
ISOLATION
- A Docker container bundles all the required libraries, making Docker containers totally independent of the underlying operating system.
- A Docker container defines a predictable environments isolated from other applications.
PERFORMANCE
- Lightweight images without operating system have much smaller footprints than virtual machines or dedicated servers and can take advantage of existent hardware.
SCALABILITY / REPRODUCIBILITY
- Environments remains more consistent in Docker; images are easily versioned, making extremely easy the test, roll back and deploy.
- Straightforward maintenance.
RUN ANYWHERE
- Docker images are free of environmental limitations, and that makes any deployment consistent, portable and scalable. Containers have the added benefit of running anywhere, providing it is targeted at the OS (Win, Mac OS, Linux, VMs, On-prem, in Public Cloud).
Train any AI model, build in any AI framework, use any data source
WIDE RANGE OF
AGGREGATORS
There are a multitude of aggregation processes, but not all of them are suitable for all models or paradigms neither offer security in untrusted environments or achieve the best accuracy in certain scenarios. A platform must offer a wide range of aggregators with which to approach learning processes of the most diverse nature and ensure that the best result is obtained from the existing model and available data.
Specific aggregators have been designed for Vertical Federated Learning (VFL). One of them is the concatenation aggregator, which, on the one hand, allows to unlock the potential of heterogeneous data; on the pother hand, assisted by Differential Privacy (DP) or Secure Multi-Party Computation (SMPC) preserves data privacy.
The federated architecture itself naturally protect data privacy. However, Adversarial Attacks may be performed to retrieve some information anyhow. Robust aggregators are designed to defend against such attacks, thus enforcing the privacy-resiliency of the federated architecture.
ROBUST
AGGREGATORS
Distributed environments architectures for Federated Learning (FL) add vulnerabilities, such as the ability to defend against adversary attacks. In classical machine learning, the majority of adversary attacks can be prevented through data sanitation techniques, where the data is inspected by cleaning it of possible contaminated data. Clearly the constraint that training data never leaves the clients in FL makes it impossible to apply these defense mechanisms and leaves a small margin of maneuver. For this reason, the development of ad-hoc defense mechanisms for FL is crucial.
Robust aggregators provide defenses against poisoning attacks (backdoor and byzantine), avoiding sample missclassification, label flipping, the introduction of samples out of the training distribution and random model updates.
DIFFERENTIAL PRIVACY
ON TOP OF EVERYTHING
Differential Privacy is a rigorous mathematical definition of privacy. Differential privacy is concerned with the idea that the information about a process and its conclussions should remain private when interchanged among different systems.
Behaving like a one-way black box, any given input provides an irreversible output that avoid data leakage.
Differential Privact is a defense against inference attacks, a subset of adversarial attacks and, along with the use of robust aggregators, provides a complete set of defenses to the distributed platform.
Regulatory
Compliance
Sherpa.ai’s Federated Learning approach allows organizations to realise the full potential of data without sharing it.
With Sherpa.ai's plataform, all computations and model training occur in the data owner's environment. This massively reduces the risk of data breaches as well as the compliance burden. Data is never shared, therefore compliance with regulations like GDPR, HIPAA HITECH or CCPA is ensured.
This innovative approach unlocks the potential of data that is currently being underutilised due to existing regulation limiting data sharing but opens up new ways of collaboration between organizations with common problems that don't currently share data for competitive reasons.
Privacy & Data Protection
Data privacy is a fundamental ethical value at Sherpa.ai.
Our platform complies with all current regulations on Data Protection (GDPR) and is in line with the European Commission regulatory framework proposal on Artificial Intelligence.
Security
Information security is a top priority at Sherpa.ai.
We believe that security must comply with quality standards and with all regulations in this regard. For this reason, we are certified in the ISO-27.001 data security standard and our platform has won the CogX 2021 awards for its Outstanding Contribution to Technology Regulation and has been a finalist as Best Solution for Privacy and Data Protection.
CONTACT SHERPA.AI
Maximize the value of data and AI with Sherpa.ai’s Privacy-Enhancing solutions
Contact us
