ProgLearn
ProgLearn copied to clipboard
NeuroData's package for exploring and using progressive learning algorithms
#39 My issue is about adjusting Proglearn so we can do scene segmentation after flattening the images. #### Reproducing code example: ``` def load_images(flatten_imgs): if flatten_imgs: X = np.array([cv2.imread(imgpath).flatten() for...
### Background A loss function suitable for unsupervised learning would contribute to extending ProgLearn's training domain to unsupervised settings in which labeled data are not available. The unsupervised contrastive loss...
### Background Currently, the progressive learning network transformer is learned as a byproduct of optimizing the softmax objective loss for classification accuracy. Contrastive loss ([reference 1](https://arxiv.org/abs/2004.11362), [reference 2](https://arxiv.org/pdf/2002.05709.pdf)) explicitly learns...
**Is your feature request related to a problem? Please describe.** We would like to devise a Reinforcement approach that leverages progressive learning to improve its in-task predictions in mapping states...
**Reference issue** #520 **Type of change** This contains the loss function and a basic ResNet50 demo demonstrating training and decreasing loss on CIFAR-10. **What does this implement/fix?** This demonstrates that...
#### Reference issue [#426](https://github.com/neurodata/ProgLearn/issues/426) #### Type of change Implementing supervised contrastive loss Adding plotting script to compare accuracies and transfer efficiencies #### What does this implement/fix? Implementing contrastive loss explicitly...
#### Reference issue Fixes issues #79, #428, #443, #446 #### Type of change Feature request #### What does this implement/fix? Adds a jupyter notebook and associated function file that demonstrate...
#### Reference issue #### Type of change Draft of SPORF. Contains cythonization, parameters equivalent to RerF, and various bug fixes from previous python-only iteration. Still needs work! #### What does...
@jdey4 we probably should use only one of the names for all classes & functions. If we try to follow `sklearn` api, would `n_estimators` be better?