Multi-Language Model-Based Filtering and Labeling of Datasets

The USPTO published a patent application (US20260100887A1) for Cisco Technology, Inc. covering a method that uses both N-gram models and large language models to make predictions about telemetry data from computer network entities. The device applies labels to the telemetry data based on both prediction outputs and provides these labels to train predictive maintenance models.

For technology companies and network infrastructure providers, this patent represents potential licensing opportunities or considerations around AI-driven network management solutions. The dual-model approach combining statistical (N-gram) and neural (LLM) methods for data labeling suggests a layered methodology for network predictive maintenance that may influence future product development strategies in the telecommunications and networking sector.