Nano Dimension has patented a method to mimic pre-trained target models without access to the original training dataset. The technology involves sending random input data to a remote device, receiving corresponding output data, and training a new model to replicate the target model’s behavior. The new model can also remove specific correlations. GlobalData’s report on Nano Dimension gives a 360-degree view of the company including its patenting strategy. Buy the report here.
Smarter leaders trust GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
According to GlobalData’s company profile on Nano Dimension, Spacecraft 3D Printing was a key innovation area identified from patents. Nano Dimension's grant share as of February 2024 was 40%. Grant share is based on the ratio of number of grants to total number of patents.
Mimicking pre-trained target model without original data access
A recently granted patent (Publication Number: US11907854B2) outlines a method for mimicking a pre-trained target model at a device without direct access to the original model or its training dataset. The method involves sending random input data to a remote device to probe the pre-trained model, generating a random probe training dataset based on the input-output data obtained, and training a new model to replicate the output data of the pre-trained model. The new model is further refined by removing desired correlations between input and output through the addition of random correlations to the training dataset.
Additionally, the patent describes a system for implementing this method, utilizing processors to send and receive data, generate training datasets, train new models, and adjust correlations. The system also includes provisions for adding new data, omitting specific data categories, re-training models, and selecting input data types based on target model outputs. The system is designed to train new models over multiple epochs, simplify model structures, and generate synthetic training samples for neural networks. Furthermore, the system can execute the new model in a run-time phase, allowing for real-time data input and output generation. Overall, the patented method and system offer a novel approach to model replication and refinement without direct access to the original model or training data.
To know more about GlobalData’s detailed insights on Nano Dimension, buy the report here.
Data Insights
From
The gold standard of business intelligence.
Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.
GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.
GlobalData Patent Analytics tracks bibliographic data, legal events data, point in time patent ownerships, and backward and forward citations from global patenting offices. Textual analysis and official patent classifications are used to group patents into key thematic areas and link them to specific companies across the world’s largest industries.
