A probabilistic constrained clustering for transfer learning and image category discovery
This work addresses scalability problems in deep clustering for computer vision tasks, offering an incremental improvement over prior methods.
The paper tackles the scalability issues of neural network-based constrained clustering for transfer learning and image category discovery by introducing a novel formulation that minimizes negative log-likelihood with pairwise constraints, resulting in improved scalability and performance without hyper-parameters.
Neural network-based clustering has recently gained popularity, and in particular a constrained clustering formulation has been proposed to perform transfer learning and image category discovery using deep learning. The core idea is to formulate a clustering objective with pairwise constraints that can be used to train a deep clustering network; therefore the cluster assignments and their underlying feature representations are jointly optimized end-to-end. In this work, we provide a novel clustering formulation to address scalability issues of previous work in terms of optimizing deeper networks and larger amounts of categories. The proposed objective directly minimizes the negative log-likelihood of cluster assignment with respect to the pairwise constraints, has no hyper-parameters, and demonstrates improved scalability and performance on both supervised learning and unsupervised transfer learning.