NoisyTwins: Class-Consistent and Diverse Image Generation through StyleGANs

NoisyTwins is a self-supervised regularization scheme for StyleGANs, which helps in alleviating mode collapse and leads to consistent conditional fine-grained image generation.

H. Rangwani*, L. Bansal*, K. Sharma, T. Karmali, V. Jampani, R.V. Babu

Conference on Computer Vision and Pattern Recognition, CVPR'23

pdf / project page / code (github)

Strata-NeRF: Neural Radiance Fields for Stratified Scenes

Strata-NeRF is a framework to model 3D scenes with various levels, using Vector Quantized Latents.

A. Dhiman, R Srinath, H. Rangwani, R. Parihar, L R Boregowda, S. Sridhar, R.V. Babu

International Conference on Computer Vision, ICCV'23

pdf / project page / code (github)

Improving GANs for Long-Tailed Data through Group Spectral Regularization

Group Spectral Regularization for alleviating mode collapse in GANs particulary for long-tailed data.

H. Rangwani, N. Jaswani, T. Karmali, V. Jampani, R.V. Babu

European Conference on Computer Vision, ECCV'22

pdf / project page / code (github)

Hierarchical Semantic Regularization of Latent Spaces in StyleGANs

Hierarchial Regularization of GAN features which enhances the smoothness in latent space

T. Karmali, R. Parihar, S. Agarwal, H. Rangwani, V. Jampani, M. Singh, R.V. Babu

European Conference on Computer Vision, ECCV'22

pdf / project page

A Closer Look at Smoothness in Domain Adversarial Training

Smooth Minima with respect to task loss leads to effective generalization on the target domain.

H. Rangwani*, S. K. Aithal*, M. Mishra, A. Jain, R. V. Babu

International Conference on Machine Learning, ICML'22

pdf / code (github) / video

Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data

Loss Re-weighting leads to convergence to saddle points, we propose high rho SAM for escaping saddles and improving generalization.

H. Rangwani*, S. K. Aithal*, M. Mishra, R. V. Babu

Neural Information Processing Systems, NeurIPS'22

pdf / code

Cost-Sensitive Self-Training for Optimizing Non-Decomposable Metrics

Consistency Regularization and Thresholding mechanism for optimizing non-decomposable objectives in a Semi-Supervised Learning Setup

H. Rangwani*, S. Ramasubramanian*, S. Takemori*, K. Takashi, Y. Umeda, R. V. Babu

Neural Information Processing Systems, NeurIPS'22

pdf / code

Class Balancing GAN with a Classifier in the Loop (CBGAN)

Pre-trained classifier based regularizer to ensure diverse and class balanced generation from an unconditional GAN.

H. Rangwani, K. R. Mopuri, R. V. Babu

Uncertainity in Artifical Intelligence, UAI'21

pdf / code (github)

S3VAADA: Submodular Subset Selection for Virtual Adversarial Active Domain Adaptation

Subset selection criterion and virtual adversarial adaptation technique for active learning for domain adaptation.

H. Rangwani, A. Jain*, S. K. Aithal*, R.V. Babu

Internation Conference on Computer Vision, ICCV'21

pdf / project page / code (github)