tinyML Talks: on-device model fine-tuning for industrial anomaly detection applications

On-device model fine-tuning for industrial anomaly detection applications

Konstantin Meshcheriakov, Solution Architect, Klika Tech delivered a tinyML Talks webcast for the tinyML Foundation in March 2022 that explored lifelong machine learning as an advanced paradigm for improving data management in constantly changing environments. See the recording for more about how applications like anomaly detection in industrial environments can be improved with on-device fine-tuning and pre-trained neural networks that can adapt to new data. See how efficient on-device learning can be done with a small memory footprint allowing models to run inference and continuously fine-tune newly collected data.
See how Klika Tech solutions can improve the flexibility of ML models and avoid common issues with continuous training, such as catastrophic forgetting. This presentation documents the process of moving an AWS cloud-based anomaly detection application to an MCU in the same time decreasing infrastructure costs and simplifying the operational efforts.