Generalization Bounds for Few-Shot Transfer Learning with Pretrained
Classifiers
- URL: http://arxiv.org/abs/2212.12532v2
- Date: Sun, 16 Jul 2023 23:41:07 GMT
- Title: Generalization Bounds for Few-Shot Transfer Learning with Pretrained
Classifiers
- Authors: Tomer Galanti, Andr\'as Gy\"orgy, Marcus Hutter
- Abstract summary: We study the ability of foundation models to learn representations for classification that are transferable to new, unseen classes.
We show that the few-shot error of the learned feature map on new classes is small in case of class-feature-variability collapse.
- Score: 26.844410679685424
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: We study the ability of foundation models to learn representations for
classification that are transferable to new, unseen classes. Recent results in
the literature show that representations learned by a single classifier over
many classes are competitive on few-shot learning problems with representations
learned by special-purpose algorithms designed for such problems. We offer a
theoretical explanation for this behavior based on the recently discovered
phenomenon of class-feature-variability collapse, that is, that during the
training of deep classification networks the feature embeddings of samples
belonging to the same class tend to concentrate around their class means. More
specifically, we show that the few-shot error of the learned feature map on new
classes (defined as the classification error of the nearest class-center
classifier using centers learned from a small number of random samples from
each new class) is small in case of class-feature-variability collapse, under
the assumption that the classes are selected independently from a fixed
distribution. This suggests that foundation models can provide feature maps
that are transferable to new downstream tasks, even with very few samples; to
our knowledge, this is the first performance bound for transfer-learning that
is non-vacuous in the few-shot setting.
Related papers
- Covariance-based Space Regularization for Few-shot Class Incremental Learning [25.435192867105552]
Few-shot Class Incremental Learning (FSCIL) requires the model to continually learn new classes with limited labeled data.
Due to the limited data in incremental sessions, models are prone to overfitting new classes and suffering catastrophic forgetting of base classes.
Recent advancements resort to prototype-based approaches to constrain the base class distribution and learn discriminative representations of new classes.
arXiv Detail & Related papers (2024-11-02T08:03:04Z) - Few-Shot Class-Incremental Learning via Training-Free Prototype
Calibration [67.69532794049445]
We find a tendency for existing methods to misclassify the samples of new classes into base classes, which leads to the poor performance of new classes.
We propose a simple yet effective Training-frEE calibratioN (TEEN) strategy to enhance the discriminability of new classes.
arXiv Detail & Related papers (2023-12-08T18:24:08Z) - Evidential Deep Learning for Class-Incremental Semantic Segmentation [15.563703446465823]
Class-Incremental Learning is a challenging problem in machine learning that aims to extend previously trained neural networks with new classes.
In this paper, we address the problem of how to model unlabeled classes while avoiding spurious feature clustering of future uncorrelated classes.
Our method factorizes the problem into a separate foreground class probability, calculated by the expected value of the Dirichlet distribution, and an unknown class (background) probability corresponding to the uncertainty of the estimate.
arXiv Detail & Related papers (2022-12-06T10:13:30Z) - GMM-IL: Image Classification using Incrementally Learnt, Independent
Probabilistic Models for Small Sample Sizes [0.4511923587827301]
We present a novel two stage architecture which couples visual feature learning with probabilistic models to represent each class.
We outperform a benchmark of an equivalent network with a Softmax head, obtaining increased accuracy for sample sizes smaller than 12 and increased weighted F1 score for 3 imbalanced class profiles.
arXiv Detail & Related papers (2022-12-01T15:19:42Z) - On the Role of Neural Collapse in Transfer Learning [29.972063833424215]
Recent results show that representations learned by a single classifier over many classes are competitive on few-shot learning problems.
We show that neural collapse generalizes to new samples from the training classes, and -- more importantly -- to new classes as well.
arXiv Detail & Related papers (2021-12-30T16:36:26Z) - Long-tail Recognition via Compositional Knowledge Transfer [60.03764547406601]
We introduce a novel strategy for long-tail recognition that addresses the tail classes' few-shot problem.
Our objective is to transfer knowledge acquired from information-rich common classes to semantically similar, and yet data-hungry, rare classes.
Experiments show that our approach can achieve significant performance boosts on rare classes while maintaining robust common class performance.
arXiv Detail & Related papers (2021-12-13T15:48:59Z) - Learning Debiased and Disentangled Representations for Semantic
Segmentation [52.35766945827972]
We propose a model-agnostic and training scheme for semantic segmentation.
By randomly eliminating certain class information in each training iteration, we effectively reduce feature dependencies among classes.
Models trained with our approach demonstrate strong results on multiple semantic segmentation benchmarks.
arXiv Detail & Related papers (2021-10-31T16:15:09Z) - Theoretical Insights Into Multiclass Classification: A High-dimensional
Asymptotic View [82.80085730891126]
We provide the first modernally precise analysis of linear multiclass classification.
Our analysis reveals that the classification accuracy is highly distribution-dependent.
The insights gained may pave the way for a precise understanding of other classification algorithms.
arXiv Detail & Related papers (2020-11-16T05:17:29Z) - Learning Adaptive Embedding Considering Incremental Class [55.21855842960139]
Class-Incremental Learning (CIL) aims to train a reliable model with the streaming data, which emerges unknown classes sequentially.
Different from traditional closed set learning, CIL has two main challenges: 1) Novel class detection.
After the novel classes are detected, the model needs to be updated without re-training using entire previous data.
arXiv Detail & Related papers (2020-08-31T04:11:24Z) - Few-Shot Learning with Intra-Class Knowledge Transfer [100.87659529592223]
We consider the few-shot classification task with an unbalanced dataset.
Recent works have proposed to solve this task by augmenting the training data of the few-shot classes using generative models.
We propose to leverage the intra-class knowledge from the neighbor many-shot classes with the intuition that neighbor classes share similar statistical information.
arXiv Detail & Related papers (2020-08-22T18:15:38Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.