Abstract
Artificial intelligence (AI) has become a valued technology in many companies. At the same time, a substantial potential for utilizing AI across company borders has remained largely untapped. An inhibiting factor concerns disclosure of data to external parties, which raises legitimate concerns about intellectual property rights, privacy issues, and cybersecurity risks. Combining federated learning with domain adaptation can provide a solution to this problem by enabling effective cross-company AI without data disclosure. In this Viewpoint, we discuss the use, value, and implications of this approach in a cross-company setting.
Introduction
Artificial Intelligence thrives on data, but in the industrial sector, the most valuable data—representing rare system failures or critical outages—is often scarce for any single company. While the obvious solution is for companies to pool their data, two major hurdles stand in the way: privacy concerns and domain shift. Companies are rightfully reluctant to share raw data that might reveal trade secrets or proprietary settings. Furthermore, even if data is shared, a model trained on one company’s machinery might fail at another due to different configurations or environments. This discrepancy in data distribution, known as "domain shift," has left an estimated $13 trillion of industrial AI potential untapped.
The Federated Transfer Learning Framework
To unlock this potential, this paper proposes a hybrid approach called Federated Transfer Learning, which merges two distinct AI paradigms to enable collaboration without data disclosure.
Federated Learning (FL)
Instead of moving data to a central server, the model is moved to the data. Companies train local models on their own servers and only share model parameters (such as gradients or weights) with a central aggregator or each other. This ensures raw data never leaves company borders, protecting intellectual property and privacy.
Domain Adaptation (DA)
To solve the "domain shift" problem, the framework utilizes DA techniques to learn "invariant representations." The goal is to extract features that are useful for the task (e.g., predicting a fault) but indistinguishable between different companies' datasets. This ensures that the collaborative model is not biased toward the company with the most data or the most "standard" operating conditions.
Strategic and Economic Implications
This framework shifts the focus from traditional horizontal collaboration (millions of mobile devices) to "vertical" or "cross-silo" collaboration involving a smaller number of entities with highly specialized data. This opens the door for several advancements:
- AI-as-a-Service: New business models can emerge where third-party orchestrators provide the federated infrastructure, allowing small and medium-sized enterprises (SMEs) to benefit from "population-wide" insights they could never generate alone.
- Systems Thinking: The challenge is framed not just as a technical one, but as a "service systems engineering" problem, requiring a systematic understanding of how value is co-created across different stakeholders.
- Decentralization: The evolution from centralized coordinators to decentralized federated learning—potentially utilizing Distributed Ledger Technology (DLT)—can enable peer-to-peer collaboration without a single point of failure.
Challenges and Future Outlook
While promising, the roadmap for cross-company AI faces several remaining hurdles:
- Privacy and Security: Standard FL does not offer a 100% guarantee of privacy; information can sometimes be reverse-engineered from gradient updates. Future work must incorporate differential privacy, cryptography, and defenses against "causative attacks" where malicious participants corrupt the model.
- Economic Incentives: There is a critical need for fair value-capture distribution models. If one company provides higher-quality data than another, the system must quantify and reward that contribution appropriately.
- Standardization: Large-scale implementation requires consistent industry standards for handling data heterogeneity and continuous learning as industrial systems evolve over time.
Conclusion
By combining Federated Learning with Domain Adaptation, companies can overcome the dual barriers of data privacy and domain shift. This architectural roadmap provides a path toward a collaborative industrial ecosystem where AI insights are shared across borders without compromising the security or proprietary interests of individual participants.
Summary updated