Information Processing Device Classifier Generation

The patent extracts the classifier generation process from the device manufacturing process. Instead of pre-installing classifiers during device production, the system allows classifiers to be generated separately using machine learning on training data, then installed or accessed by end users as needed. This separation eliminates the need for expensive preconfiguration during manufacturing while maintaining recognition functionality.

The patent prepares training data and classifier generation resources in advance, but separates the actual classifier generation from device manufacturing. The machine learning models and training datasets are prepared beforehand, allowing classifiers to be generated on-demand without increasing production costs. This preliminary preparation enables flexible, post-manufacturing customization.

The patent performs classifier generation in advance using available training data, storing the generated classifiers for later use. This preliminary generation avoids the need for users to wait for classifier training when they need recognition functionality. The system pre-processes learning data and generates classifiers before actual recognition tasks begin.

The patent segments the recognition system into separate components: training data preparation, classifier generation, and recognition execution. This segmentation allows classifier generation to be performed independently and stored, separating the time-consuming training phase from the fast recognition phase. Multiple classifiers can be generated in parallel using different training datasets.