Image embeddings via deep learning and adaptive batch normalization

One or more non-transitory computer readable media storing instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of:

• inputting a first batch of images into a first trained machine learning model that includes a plurality of layers, wherein the first batch of images has been sampled from a first plurality of images generated via a first experiment;
• for at least one layer included in the plurality of layers:
• computing a first set of statistics associated with a first plurality of outputs generated by the at least one layer based on the first batch of images;
• and
• normalizing the first plurality of outputs based on the first set of statistics to generate a first plurality of normalized outputs;
• and
• generating a first plurality of embeddings for the first batch of images based on the first plurality of normalized outputs.

A system, comprising:

• a memory that stores instructions, and
• a processor that is coupled to the memory and, when executing the instructions, is configured to:
• input a first batch of images into a first machine learning model that includes a plurality of layers, wherein the first batch of images has been sampled from a first plurality of images generated via a first experiment;
• for at least one layer included in the plurality of layers:
• compute a first set of statistics associated with a first plurality of outputs generated by the at least one layer based on the first batch of images;
• and
• normalize the first plurality of outputs based on the first set of statistics to generate a first plurality of normalized outputs;
• and
• update a plurality of parameters for each layer included in the plurality of layers based on a first set of predictions generated by the first machine learning model from the first batch of images and the first plurality of normalized outputs.