According to GlobalData’s company profile on ANZ Group, AI-assisted security risk prediction was a key innovation area identified from patents. ANZ Group's grant share as of June 2023 was 1%. Grant share is based on the ratio of number of grants to total number of patents.

Method for selecting hybrid variables using machine learning

Source: United States Patent and Trademark Office(USPTO). Credit: ANZ Group Holdings Ltd

A recently filed patent (Publication Number: US20230146635A1) describes a method for generating a machine learning model. The method involves sampling interaction effect structures and hybrid variables from a multivariable dataset. The lift value of each sampled hybrid variable is calculated and compared to a threshold lift criteria. If the lift value exceeds the threshold, the hybrid variable is labeled. A machine learning model is then trained using the labeled hybrid variables to predict the likelihood of a hybrid variable having a lift that exceeds the threshold. The trained model is applied to each hybrid variable within the sampled interaction effect structures to determine the likelihood value. Only hybrid variables with a likelihood value that exceeds a decision criteria are retained.

The method also includes determining whether the number of retained hybrid variables exceeds a predetermined threshold. If it does not, additional interaction effect structures are sampled, and the method is repeated. A discriminatory strength statistic is calculated for each retained hybrid variable, and hybrid variables that do not meet a discriminatory strength statistic decision criteria are discarded. The retained hybrid variables can be sorted based on the discriminatory strength statistic and the predicted lift likelihood value.

The multivariable dataset used in the method includes dependent variables and independent variables. The dependent variables are labeled variables, and the dataset can be partitioned based on the labeled dependent variables to create multiple partitioned datasets. Discriminatory strength statistics are calculated for each variable in the partitioned datasets, as well as for each sampled hybrid variable.

Moment statistics are calculated for each variable and hybrid variable, and the moment statistics calculated for the selected variables are sourced. The method also involves creating a variable moments dataset and storing the moment statistics of each variable within it. A moments dataset is created to store the moment statistics of each hybrid variable alongside the moment statistics of the selected variables for the corresponding hybrid variables.

The patent also describes a system for selecting hybrid variables, which includes a processor and memory storing program code. The processor executes the program code to sample interaction effect structures and hybrid variables, calculate lift values, label hybrid variables, train a machine learning model, apply the model, and retain hybrid variables with a likelihood value that exceeds a decision criteria.

Overall, this patent presents a method and system for generating a machine learning model using hybrid variables and interaction effect structures from a multivariable dataset. The method includes various steps such as sampling, calculating lift values, labeling, training, and retaining hybrid variables based on likelihood values. The system utilizes a processor and memory to execute the method.

To know more about GlobalData’s detailed insights on ANZ Group, buy the report here.

Premium 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.