Federated learning and time series are elementary in IBM Watson

Federated learning and time series are elementary in IBM Watson
IBM image via shutterstock

IBM is making updates to Watson AI technologies for business, including the addition of support for federated learning and time series capabilities in Watson Studio. Big Blue said the changes are designed to help organisations build trustworthy AI solutions.

According to IBM, the new capabilities expand on Watson tools built to help businesses govern and explain AI-led decisions, increase insight accuracy, mitigate risks and meet their privacy and compliance requirements, which might otherwise prove to be barriers to trust in data and AI models.

Among the updates is federated learning capability in Watson Studio, IBM’s data science platform. This is currently only a technical preview, but will enable organisations to train AI models on data sources that would previously have been separate silos. Such silos may exist where data cannot or should not be moved due to reasons such as data privacy, secrecy, regulatory compliance, or simply the sheer volume of data involved.

Also added to Watson Studio is beta support for time series capabilities, to tackle the challenges of automating, analysing, and forecasting time series data commonly seen in various industries such as financial services, manufacturing, and retail. IBM said that Time Series on Watson Studio is designed to achieve accuracy on a range of datasets such as phone call data logs, weather data, travel times, or retail sales.

IBM OpenPages with Watson now includes a Data Privacy Management module designed to help businesses meet evolving data privacy challenges. By integrating with Watson Knowledge Catalog, IBM OpenPages can provide businesses with a near real-time view of how private data is being used throughout the organisation, IBM claimed.

IBM Planning Analytics with Watson is set to gain a new statistical details page to provide more transparent and easy-to-understand facts about how a forecasting prediction was generated. IBM said that businesses require transparency in the models and data used to generate predictive forecasts if they are to be used in areas such as financial, sales and supply chain planning. This feature is expected to be available in Q2 of 2021.