The United States Patent and Trademark Office (USPTO) has granted patent US12585983B2, titled "Machine learning training using ground truth approximations," to Aptiv Technologies AG. The patent describes a computer-implemented method for training machine learning models. This method involves using measurement data from a first sensor and approximations of ground truths derived from a second sensor. The training process is designed such that approximations of lower quality have a reduced impact compared to those of higher quality, aiming to improve the robustness and accuracy of the trained models.

This patent grant represents a new intellectual property asset for Aptiv Technologies AG in the field of AI and machine learning. For compliance officers, this is primarily an informational item related to IP protection in AI development. There are no immediate compliance obligations or deadlines for external entities. The patent's claims and classifications (CPC: G06N 20/00) indicate its focus on advanced machine learning techniques, particularly in areas where precise ground truth data may be difficult to obtain.