A method includes receiving a text description of a system capability request, and converting the text description into a normalized description of the system capability request. A repository is then queried, based on the normalized description and using a search algorithm, to identify multiple candidate application software units (ASUs). The candidate ASUs are displayed to a user for selection. The user-selected ASU is then deployed, either locally or to at least one remote compute device, in response to receiving the user selection. Deployment can include the user-selected candidate ASU being integrated into a local or remote software package, thus defining a modified software package that is configured to provide the system capability.

A method includes receiving a text description of a system capability request, and converting the text description into a normalized description of the system capability request. A repository is then queried, based on the normalized description and using a search algorithm, to identify multiple candidate application software units (ASUs). The candidate ASUs are displayed to a user for selection. The user-selected ASU is then deployed, either locally or to at least one remote compute device, in response to receiving the user selection. Deployment can include the user-selected candidate ASU being integrated into a local or remote software package, thus defining a modified software package that is configured to provide the system capability.

The present disclosure relates to systems and method of determining transcriptomic profile from omics imaging data. The systems and methods train machine learning methods with intrinsic and extrinsic features of a cell and/or tissue to define transcriptomic profiles of the cell and/or tissue. Applicants utilize a convolutional autoencoder to define cell subtypes from images of the cells.

Fair deep reinforcement learning is provided. A microstate of an environment and reaction of items in a plurality of microstates within the environment are observed after an agent performs an action in the environment. Semi-supervised training is utilized to determine bias weights corresponding to the action for the microstate of the environment and the reaction of the items in the plurality of microstates within the environment. The bias weights from the semi-supervised training are merged with non-bias weights using an artificial neural network. Over time, it is determined where bias is occurring in the semi-supervised training based on merging the bias weights with the non-bias weights in the artificial neural network. A deep reinforcement learning model that decreases reliance on the bias weights is generated based on determined bias to increase fairness.

An instance weighted learning (IWL) machine learning model. In one example embodiment, a method of employing an IWL machine learning model to train a classifier may include determining a quality value that should be associated with each machine learning training instance in a temporal sequence of reinforcement learning machine learning training instances, associating the corresponding determined quality value with each of the machine learning training instances, and training a classifier using each of the machine learning training instances. Each of the machine learning training instances includes a state-action pair and is weighted during the training based on its associated quality value using a weighting factor that weights different quality values differently such that the classifier learns more from a machine learning training instance with a higher quality value than from a machine learning training instance with a lower quality value.