Simon Lacoste-Julien

PopulAtion Parameter Averaging (PAPA)

Alexia Jolicoeur-Martineau

Emy Gervais

Kilian FATRAS

Yan Zhang

Ensemble methods combine the predictions of multiple models to improve performance, but they require significantly higher computation costs … (see more)at inference time. To avoid these costs, multiple neural networks can be combined into one by averaging their weights. However, this usually performs significantly worse than ensembling. Weight averaging is only beneficial when different enough to benefit from combining them, but similar enough to average well. Based on this idea, we propose PopulAtion Parameter Averaging (PAPA): a method that combines the generality of ensembling with the efficiency of weight averaging. PAPA leverages a population of diverse models (trained on different data orders, augmentations, and regularizations) while slowly pushing the weights of the networks toward the population average of the weights. We also propose PAPA variants (PAPA-all, and PAPA-2) that average weights rarely rather than continuously; all methods increase generalization, but PAPA tends to perform best. PAPA reduces the performance gap between averaging and ensembling, increasing the average accuracy of a population of models by up to 0.8% on CIFAR-10, 1.9% on CIFAR-100, and 1.6% on ImageNet when compared to training independent (non-averaged) models.

2023-04-06

ArXiv (preprint)

PopulAtion Parameter Averaging (PAPA)

Alexia Jolicoeur-Martineau

Emy Gervais

Kilian FATRAS

Yan Zhang

Ensemble methods combine the predictions of multiple models to improve performance, but they require significantly higher computation costs … (see more)at inference time. To avoid these costs, multiple neural networks can be combined into one by averaging their weights. However, this usually performs significantly worse than ensembling. Weight averaging is only beneficial when different enough to benefit from combining them, but similar enough to average well. Based on this idea, we propose PopulAtion Parameter Averaging (PAPA): a method that combines the generality of ensembling with the efficiency of weight averaging. PAPA leverages a population of diverse models (trained on different data orders, augmentations, and regularizations) while slowly pushing the weights of the networks toward the population average of the weights. We also propose PAPA variants (PAPA-all, and PAPA-2) that average weights rarely rather than continuously; all methods increase generalization, but PAPA tends to perform best. PAPA reduces the performance gap between averaging and ensembling, increasing the average accuracy of a population of models by up to 0.8% on CIFAR-10, 1.9% on CIFAR-100, and 1.6% on ImageNet when compared to training independent (non-averaged) models.

2023-04-06

ArXiv (preprint)

A Survey of Self-Supervised and Few-Shot Object Detection

Gabriel Huang

Issam Hadj Laradji

David Vazquez

Pau Rodriguez

Labeling data is often expensive and time-consuming, especially for tasks such as object detection and instance segmentation, which require … (see more)dense labeling of the image. While few-shot object detection is about training a model on novel (unseen) object classes with little data, it still requires prior training on many labeled examples of base (seen) classes. On the other hand, self-supervised methods aim at learning representations from unlabeled data which transfer well to downstream tasks such as object detection. Combining few-shot and self-supervised object detection is a promising research direction. In this survey, we review and characterize the most recent approaches on few-shot and self-supervised object detection. Then, we give our main takeaways and discuss future research directions. Project page: https://gabrielhuang.github.io/fsod-survey/.

2023-04-01

IEEE Transactions on Pattern Analysis and Machine Intelligence (published)

Can We Scale Transformers to Predict Parameters of Diverse ImageNet Models?

Boris Knyazev

DOHA HWANG

Pretraining a neural network on a large dataset is becoming a cornerstone in machine learning that is within the reach of only a few communi… (see more)ties with large-resources. We aim at an ambitious goal of democratizing pretraining. Towards that goal, we train and release a single neural network that can predict high quality ImageNet parameters of other neural networks. By using predicted parameters for initialization we are able to boost training of diverse ImageNet models available in PyTorch. When transferred to other datasets, models initialized with predicted parameters also converge faster and reach competitive final performance.

2023-03-07

ArXiv (preprint)

Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learning

Sébastien Lachapelle

Tristan Deleu

Divyat Mahajan

Quentin Bertrand

Although disentangled representations are often said to be beneficial for downstream tasks, current empirical and theoretical understanding … (see more)is limited. In this work, we provide evidence that disentangled representations coupled with sparse base-predictors improve generalization. In the context of multi-task learning, we prove a new identifiability result that provides conditions under which maximally sparse base-predictors yield disentangled representations. Motivated by this theoretical result, we propose a practical approach to learn disentangled representations based on a sparsity-promoting bi-level optimization problem. Finally, we explore a meta-learning version of this algorithm based on group Lasso multiclass SVM base-predictors, for which we derive a tractable dual formulation. It obtains competitive results on standard few-shot classification benchmarks, while each task is using only a fraction of the learned representations.

2023-01-01

ICML (published)

Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learning

Sébastien Lachapelle

Tristan Deleu

Divyat Mahajan

Quentin Bertrand

Although disentangled representations are often said to be beneficial for downstream tasks, current empirical and theoretical understanding … (see more)is limited. In this work, we provide evidence that disentangled representations coupled with sparse task-specific predictors improve generalization. In the context of multi-task learning, we prove a new identifiability result that provides conditions under which maximally sparse predictors yield disentangled representations. Motivated by this theoretical result, we propose a practical approach to learn disentangled representations based on a sparsity-promoting bi-level optimization problem. Finally, we explore a meta-learning version of this algorithm based on group Lasso multiclass SVM predictors, for which we derive a tractable dual formulation. It obtains competitive results on standard few-shot classification benchmarks, while each task is using only a fraction of the learned representations.

2023-01-01

ICML (published)

proceedings.mlr.press

CrossSplit: Mitigating Label Noise Memorization through Data Splitting

Jihye Kim

Aristide Baratin

Yan Zhang

We approach the problem of improving robustness of deep learning algorithms in the presence of label noise. Building upon existing label cor… (see more)rection and co-teaching methods, we propose a novel training procedure to mitigate the memorization of noisy labels, called CrossSplit, which uses a pair of neural networks trained on two disjoint parts of the labelled dataset. CrossSplit combines two main ingredients: (i) Cross-split label correction. The idea is that, since the model trained on one part of the data cannot memorize example-label pairs from the other part, the training labels presented to each network can be smoothly adjusted by using the predictions of its peer network; (ii) Cross-split semi-supervised training. A network trained on one part of the data also uses the unlabeled inputs of the other part. Extensive experiments on CIFAR-10, CIFAR-100, Tiny-ImageNet and mini-WebVision datasets demonstrate that our method can outperform the current state-of-the-art in a wide range of noise ratios.

2022-12-03

ArXiv (preprint)

Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learning

Sébastien Lachapelle

Tristan Deleu

Divyat Mahajan

Quentin Bertrand

Although disentangled representations are often said to be beneficial for downstream tasks, current empirical and theoretical understanding … (see more)is limited. In this work, we provide evidence that disentangled representations coupled with sparse base-predictors improve generalization. In the context of multi-task learning, we prove a new identifiability result that provides conditions under which maximally sparse base-predictors yield disentangled representations. Motivated by this theoretical result, we propose a practical approach to learn disentangled representations based on a sparsity-promoting bi-level optimization problem. Finally, we explore a meta-learning version of this algorithm based on group Lasso multiclass SVM base-predictors, for which we derive a tractable dual formulation. It obtains competitive results on standard few-shot classification benchmarks, while each task is using only a fraction of the learned representations.

2022-11-26

ArXiv (preprint)

Synergies between Disentanglement and Sparsity: Generalization and Identifiability in Multi-Task Learning

Sébastien Lachapelle

Tristan Deleu

Divyat Mahajan

Quentin Bertrand

Although disentangled representations are often said to be beneficial for downstream tasks, current empirical and theoretical understanding … (see more)is limited. In this work, we provide evidence that disentangled representations coupled with sparse base-predictors improve generalization. In the context of multi-task learning, we prove a new identifiability result that provides conditions under which maximally sparse base-predictors yield disentangled representations. Motivated by this theoretical result, we propose a practical approach to learn disentangled representations based on a sparsity-promoting bi-level optimization problem. Finally, we explore a meta-learning version of this algorithm based on group Lasso multiclass SVM base-predictors, for which we derive a tractable dual formulation. It obtains competitive results on standard few-shot classification benchmarks, while each task is using only a fraction of the learned representations.

2022-11-26

ArXiv (preprint)

SVRG meets AdaGrad: painless variance reduction

Benjamin Dubois-Taine

Sharan Vaswani

Reza Babanezhad Harikandeh

Mark Schmidt

2022-11-10

Machine Learning (published)

Controlled Sparsity via Constrained Optimization or: How I Learned to Stop Tuning Penalties and Love Constraints

Jose Gallego-Posada

Juan Ramirez

Akram Erraqabi

The performance of trained neural networks is robust to harsh levels of pruning. Coupled with the ever-growing size of deep learning models,… (see more) this observation has motivated extensive research on learning sparse models. In this work, we focus on the task of controlling the level of sparsity when performing sparse learning. Existing methods based on sparsity-inducing penalties involve expensive trial-and-error tuning of the penalty factor, thus lacking direct control of the resulting model sparsity. In response, we adopt a constrained formulation: using the gate mechanism proposed by Louizos et al. (2018), we formulate a constrained optimization problem where sparsification is guided by the training objective and the desired sparsity target in an end-to-end fashion. Experiments on CIFAR-{10, 100}, TinyImageNet, and ImageNet using WideResNet and ResNet{18, 50} models validate the effectiveness of our proposal and demonstrate that we can reliably achieve pre-determined sparsity targets without compromising on predictive performance.