Transfer learning with machine learning systems

A system comprising:

• a processor;
• and
• a memory device coupled with the processor;
• the processor configured at least to:
• receive a machine learning model;
• receive target domain training data for reprogramming the machine learning model using transfer learning;
• transform the target domain training data by performing a transformation function on the target domain training data, wherein the target domain training data is padded, wherein the transformation function includes at least a trainable parameter and a binary mask function, wherein the transformation function is used in implementing adversarial reprogramming in machine learning, wherein parameters of the machine learning model are unobservable;
• map output labels of the machine learning model to target labels associated with the target domain training data, wherein the processor is configured to map the output labels by frequency-based mapping, wherein output label prediction distribution of the target domain training data is obtained from the machine learning model before reprogramming, and a most frequent output label prediction is assigned to a corresponding dominating target label until each target label is assigned with a source label;
• and
• train the transformation function by optimizing parameter of the transformation function.

A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions readable by a device to cause the device to:

• receive a machine learning model;
• receive target domain training data for reprogramming the machine learning model using transfer learning;
• transform the target domain training data by performing a transformation function on the target domain training data, wherein the target domain training data is padded, wherein the transformation function includes at least a trainable parameter and a binary mask function, wherein the transformation function is used in implementing adversarial reprogramming in machine learning, wherein parameters of the machine learning model are unobservable;
• map output labels of the machine learning model to target labels associated with the target domain training data, wherein the device is caused to map the output labels by frequency-based mapping, wherein output label prediction distribution of the target domain training data is obtained from the machine learning model before reprogramming, and a most frequent output label prediction is assigned to a corresponding dominating target label until each target label is assigned with a source label;
• and
• train the transformation function by optimizing parameter of the transformation function.