Abstract: An approach is disclosed herein to sequence selection in a UVM environment. Generally, this approach includes a training phase for each machine learning model of a plurality of machine learning models. Each model is trained to achieve a particular target state and is rewarded when a selected action or sequence of actions causes movement that might be beneficial to achieving that target state. Once a respective model is trained, the trained model can then be used to determine which one action or sequence of actions (or ordered multiple thereof) to take to achieve the corresponding target state. Thus, by training and using a plurality of machine learning models to achieve a plurality of target states, and stimulating those machine learning models once trained, one or more actions and/or sequences of actions are generated as the selected sequences to be used to verify functionality or operation of a design under test.

Abstract: An approach is disclosed herein for test sequence processing that is applicable to machine learning model generated test sequences as disclosed herein. The test sequence processing includes classification, grouping, and filtering. The classification is generated based on the execution of the test sequences. The grouping is performed based on information captured during the classification of the test sequences. The filtering is performed on a group by group basis to remove redundant test sequences.

Abstract: The approach disclosed herein is a new approach to sequence generation in the context of validation that relies on machine learning to explore and identify ways to achieve different states. In particular, the approach uses machine learning models to identify different states and ways to transition from one state to another. Actions are selected by machine learning models as they are being trained using reinforcement learning. This online inference also is likely to result in the discovery of not yet discovered states. Each state that has been identified is then used as a target to train a respective machine learning model. As part of this process a representation of all the states and actions or sequences of actions executed to reach those states is created. This representation, the respective machine learning models, or a combination thereof can then be used to generate different test sequences.