Implicit Equivariance in Convolutional Networks
This addresses the problem of designing efficient and flexible equivariant CNNs for computer vision tasks, offering an incremental improvement over existing methods.
The paper tackled the lack of inherent equivariance in CNNs for transformations like rotations and scaling by proposing Implicitly Equivariant Networks (IEN), which uses a multi-objective loss to induce equivariance, resulting in better performance than steerable networks and enabling a 30% reduction in channels while maintaining baseline performance.
Convolutional Neural Networks(CNN) are inherently equivariant under translations, however, they do not have an equivalent embedded mechanism to handle other transformations such as rotations and change in scale. Several approaches exist that make CNNs equivariant under other transformation groups by design. Among these, steerable CNNs have been especially effective. However, these approaches require redesigning standard networks with filters mapped from combinations of predefined basis involving complex analytical functions. We experimentally demonstrate that these restrictions in the choice of basis can lead to model weights that are sub-optimal for the primary deep learning task (e.g. classification). Moreover, such hard-baked explicit formulations make it difficult to design composite networks comprising heterogeneous feature groups. To circumvent such issues, we propose Implicitly Equivariant Networks (IEN) which induce equivariance in the different layers of a standard CNN model by optimizing a multi-objective loss function that combines the primary loss with an equivariance loss term. Through experiments with VGG and ResNet models on Rot-MNIST , Rot-TinyImageNet, Scale-MNIST and STL-10 datasets, we show that IEN, even with its simple formulation, performs better than steerable networks. Also, IEN facilitates construction of heterogeneous filter groups allowing reduction in number of channels in CNNs by a factor of over 30% while maintaining performance on par with baselines. The efficacy of IEN is further validated on the hard problem of visual object tracking. We show that IEN outperforms the state-of-the-art rotation equivariant tracking method while providing faster inference speed.