Chidubem Arachie is a DAC Ph.D. student in the Department of Computer Science and his advisor is Bert Huang.
Arachie’s research interest lies in investigating new methods for weakly supervised learning. Labeled training data can be a limitation when training maching learning models and in these situations, weak supervsion can provide a suitable alternative. Weakly supervised learning trains models with unlabeled data and noisy indicators of the target distribution.
Currently, he is working on combining adversarial learning with weak supervision to train models robust to noise and error dependencies among the weak supervision signals.