Minimax curriculum learning: Machine teaching with desirable difficulties and scheduled diversity
We introduce and study minimax curriculum learning (MCL), a new method for adaptively
selecting a sequence of training subsets for a succession of stages in machine learning. The …
selecting a sequence of training subsets for a succession of stages in machine learning. The …
Superloss: A generic loss for robust curriculum learning
T Castells, P Weinzaepfel… - Advances in Neural …, 2020 - proceedings.neurips.cc
Curriculum learning is a technique to improve a model performance and generalization
based on the idea that easy samples should be presented before difficult ones during …
based on the idea that easy samples should be presented before difficult ones during …
Theory of curriculum learning, with convex loss functions
D Weinshall, D Amir - Journal of Machine Learning Research, 2020 - jmlr.org
Curriculum Learning is motivated by human cognition, where teaching often involves
gradually exposing the learner to examples in a meaningful order, from easy to hard …
gradually exposing the learner to examples in a meaningful order, from easy to hard …
On the statistical benefits of curriculum learning
Curriculum learning (CL) is a commonly used machine learning training strategy. However,
we still lack a clear theoretical understanding of CL's benefits. In this paper, we study the …
we still lack a clear theoretical understanding of CL's benefits. In this paper, we study the …
Gradient-based algorithms for machine teaching
P Wang, K Nagrecha… - Proceedings of the IEEE …, 2021 - openaccess.thecvf.com
The problem of machine teaching is considered. A new formulation is proposed under the
assumption of an optimal student, where optimality is defined in the usual machine learning …
assumption of an optimal student, where optimality is defined in the usual machine learning …
Learning a minimax optimizer: A pilot study
Solving continuous minimax optimization is of extensive practical interest, yet notoriously
unstable and difficult. This paper introduces the learning to optimize (L2O) methodology to …
unstable and difficult. This paper introduces the learning to optimize (L2O) methodology to …
Curriculum learning by transfer learning: Theory and experiments with deep networks
D Weinshall, G Cohen, D Amir - International conference on …, 2018 - proceedings.mlr.press
We provide theoretical investigation of curriculum learning in the context of stochastic
gradient descent when optimizing the convex linear regression loss. We prove that the rate …
gradient descent when optimizing the convex linear regression loss. We prove that the rate …
Curriculum learning: A survey
Training machine learning models in a meaningful order, from the easy samples to the hard
ones, using curriculum learning can provide performance improvements over the standard …
ones, using curriculum learning can provide performance improvements over the standard …
Curriculum learning
Humans and animals learn much better when the examples are not randomly presented but
organized in a meaningful order which illustrates gradually more concepts, and gradually …
organized in a meaningful order which illustrates gradually more concepts, and gradually …
On the power of curriculum learning in training deep networks
G Hacohen, D Weinshall - International conference on …, 2019 - proceedings.mlr.press
Training neural networks is traditionally done by providing a sequence of random mini-
batches sampled uniformly from the entire training data. In this work, we analyze the effect of …
batches sampled uniformly from the entire training data. In this work, we analyze the effect of …