Adaptive Curriculum Learning

Kong, Yajing; Liu, Liu; Wang, Jun; Tao, Dacheng

doi:10.1109/iccv48922.2021.00502

Cited by 15 publications

(6 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is further improved in [56] by encouraging more exploration during early phases of learning. More recently, [18] propose a curriculum which computes exponential moving averages of loss values as difficulty scores for training samples.…”

Section: Related Workmentioning

confidence: 99%

Adaptive Sample Selection for Robust Learning under Label Noise

Patel

Sastry

2023

2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) have been shown to be susceptible to memorization or overfitting in the presence of noisily-labelled data.For the problem of robust learning under such noisy data, several algorithms have been proposed. A prominent class of algorithms rely on sample selection strategies wherein, essentially, a fraction of samples with loss values below a certain threshold are selected for training. These algorithms are sensitive to such thresholds, and it is difficult to fix or learn these thresholds. Often, these algorithms also require information such as label noise rates which are typically unavailable in practice. In this paper, we propose an adaptive sample selection strategy that relies only on batch statistics of a given mini-batch to provide robustness against label noise. The algorithm does not have any additional hyperparameters for sample selection, does not need any information on noise rates and does not need access to separate data with clean labels. We empirically demonstrate the effectiveness of our algorithm on benchmark datasets. 1

show abstract

Section: Related Workmentioning

confidence: 99%

Adaptive Sample Selection for Robust Learning under Label Noise

Patel

Sastry

2023

2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

View full text Add to dashboard Cite

show abstract

“…We use scores based on pretrained models (Zhang and Bansal, 2019; to compute difficulty. Kong et al (2021) and Cai et al (2020) demonstrate that an adaptive curriculum can improve convergence for image classification and neural response generation respectively. We examine adaptive CL for commonsense reasoning.…”

Section: Related Workmentioning

confidence: 99%

“…When the optimal curriculum for a student is not known in advance, a teacher usually draws up a curriculum based on past teaching experience and then adjusts with the learning progress of the student. Accordingly, Kong et al (2021) propose initializing the curriculum using the difficulty score obtained from the teacher model and then adapt the score to the current state of the learner model (see Fig. 1).…”

Section: Adaptive Curriculum Learningmentioning

confidence: 99%

“…Hence, we also investigate adaptive curriculum learning for commonsense reasoning tasks. The initial data order imposed by the teacher model is updated at regular intervals during training by taking the learner model's current state into account (Kong et al, 2021). Importantly, we propose to reverse the ranking to a difficulty-to-easy curriculum in ACL, in order to reinforce feedback from the hardto-learn data points, which has been shown to be beneficial for generalization .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On Curriculum Learning for Commonsense Reasoning

Maharana¹,

Bansal²

2022

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

View full text Add to dashboard Cite

Commonsense reasoning tasks follow a standard paradigm of finetuning pretrained language models on the target task data, where samples are introduced to the model in a random order during training. However, recent research suggests that data order can have a significant impact on the performance of finetuned models for natural language understanding. Hence, we examine the effect of a humanlike easy-to-difficult curriculum during finetuning of language models for commonsense reasoning tasks. We use paced curriculum learning to rank data and sample training mini-batches with increasing levels of difficulty from the ranked dataset during finetuning. Further, we investigate the effect of an adaptive curriculum, i.e., the data ranking is dynamically updated during training based on the current state of the learner model. We use a teacher model to measure difficulty of each sample and experiment with three measures based on question answering probability, variability and out-ofdistribution. To understand the effectiveness of curriculum learning in various scenarios, we apply it on full model fine-tuning as well as parameter-efficient prompt-tuning settings. Our results show that fixed as well as adaptive curriculum learning significantly improve performance for five commonsense reasoning tasks, i.e., SocialIQA, CosmosQA, CODAH, HellaSwag, WinoGrande in both tuning settings. Further, we find that prioritizing the difficult samples in the tail end of training improves generalization to unseen in-domain data as well as out-of-domain data. Our work provides evidence and encourages research into curriculum learning for commonsense reasoning. 1 .

show abstract

“…The latest literature (Delange et al 2021) proposed a series of guidelines for continual learning methods to be applicable in practice: i) good performance and less forgetting on previous tasks; ii) no oracle of task identifiers at inference time; and iii) bounded memory footprint throughout the entire training phase. Unfortunately, most of the exiting methods fail to satisfy all these guidelines mentioned above (Chaudhry et al 2021;Wang et al 2020;Lopez-Paz et al 2017;Wei et al 2021;Riemer et al 2019;Kong et al 2021;Schwarz et al 2018;Zhu et al 2020).…”

Section: Introductionmentioning

confidence: 99%

Continual Learning through Retrieval and Imagination

Wang

Liu

Duan

et al. 2022

AAAI

Self Cite

View full text Add to dashboard Cite

Continual learning is an intellectual ability of artificial agents to learn new streaming labels from sequential data. The main impediment to continual learning is catastrophic forgetting, a severe performance degradation on previously learned tasks. Although simply replaying all previous data or continuously adding the model parameters could alleviate the issue, it is impractical in real-world applications due to the limited available resources. Inspired by the mechanism of the human brain to deepen its past impression, we propose a novel framework, Deep Retrieval and Imagination (DRI), which consists of two components: 1) an embedding network that constructs a unified embedding space without adding model parameters on the arrival of new tasks; and 2) a generative model to produce additional (imaginary) data based on the limited memory. By retrieving the past experiences and corresponding imaginary data, DRI distills knowledge and rebalances the embedding space to further mitigate forgetting. Theoretical analysis demonstrates that DRI can reduce the loss approximation error and improve the robustness through retrieval and imagination, bringing better generalizability to the network. Extensive experiments show that DRI performs significantly better than the existing state-of-the-art continual learning methods and effectively alleviates catastrophic forgetting.

show abstract

Adaptive Curriculum Learning

Cited by 15 publications

References 24 publications

Adaptive Sample Selection for Robust Learning under Label Noise

Adaptive Sample Selection for Robust Learning under Label Noise

On Curriculum Learning for Commonsense Reasoning

Continual Learning through Retrieval and Imagination

Contact Info

Product

Resources

About