The new update enables real-time machine learning for signal analysis and more, using modular platform tools for research applications.
Liquid Instruments has incorporated artificial intelligence (AI) into its Moku platform. This update introduces the Moku Neural Network, adding the capability to execute machine learning (ML) models for applications such as signal analysis, denoising, sensor conditioning, and more.
Moku uses the computational strength of field-programmable gate arrays (FPGAs). This technology supports a range of tools, from lab essentials like oscilloscopes to devices like lock-in amplifiers, all integrated into a single platform. The FPGA’s versatility and processing capabilities, combined with integration with other Moku tools, allow users to develop and deploy their own artificial neural networks using Python on the Moku:Pro device.
The architecture of the Moku Neural Network features input, hidden, and output layers with customizable activation functions. It supports applications with features like sub-microsecond latency, up to five fully connected layers with as many as 100 neurons each, and the ability to handle up to four input channels in parallel or process time-series data in serial mode with up to four outputs—all in real time.
To assist users in leveraging the Moku technology, Liquid Instruments introduces AI Help, a GenAI tool designed to quickly address product inquiries, simplifying the use and maximization of Moku’s capabilities. Like many GenAI tools, it is expected to receive ongoing enhancements and broader functional applications.
“The Moku Neural Network is designed to enable scientists and engineers to easily integrate machine learning into their experiments,” said Daniel Shaddock, co-founder and CEO of Liquid Instruments. “Moku has always offered flexible and easy-to-use tools for advanced R&D. This latest release demonstrates the platform’s extensibility, adding new AI/ML capabilities through a software update.”
The company provides users with a suite of test instruments that can be reconfigured. Regular software updates, machine learning capabilities, and the ability to integrate new technologies ensure that the Moku platform evolves with its users. This approach aims to transform the cost and effort required to move from concept to implementation in research and development applications.
