Swisscom has been granted a patent for methods and systems for neural architecture search. The technology involves determining a preferred model for a task by obtaining a computational graph, training subgraphs sequentially, and updating shared weightings based on importance indications. The neural network can be configured for tasks like natural language processing and image recognition. GlobalData’s report on Swisscom 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 Swisscom, Mobile payments was a key innovation area identified from patents. Swisscom's grant share as of May 2024 was 65%. Grant share is based on the ratio of number of grants to total number of patents.
Neural architecture search method for training neural networks
A recently granted patent (Publication Number: US12001957B2) outlines a method for determining a preferred model for performing a selected task within a system comprising one or more processing circuits. The method involves utilizing a computational graph with nodes, edges, and weightings to define different models, updating the weightings of these models based on training, and selecting the preferred model through an analysis of the trained models. The system then configures a neural network based on the preferred model to perform tasks such as natural language processing, image recognition, classification, data processing, and generating search results.
Furthermore, the patent details the process of determining the importance of shared weightings between models, initializing weightings of the second model based on the trained first model, and identifying the preferred model through reinforcement learning or evolutionary learning. The method involves operating a controller based on a reward system that considers the goodness of fit and complexity of the model, repeating steps for different suggested models until a threshold is reached, and selecting the best-performing suggested model as the architecture for the neural network. The patent also emphasizes the repetition of updating weightings and identifying the preferred model for a selected number of times to optimize model performance.
To know more about GlobalData’s detailed insights on Swisscom, 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.
