Implementation of several neural network models for multiple instance learning (MIL). Models include Multiple Instance Net (MINet), MINet with Deep Supervision(MINetDS), and MINet with Residual Connections (MINetRC). Code includes several pooling methods including mean, max, and log-sum-exp (LSE).
Keywords: multiple instance learning, MINet, MINetDS, MINetRC, neural networks, classification, machine learning, Tensorflow, Keras