Rebalancing Curriculum Learning via In-Batch Difficulty Reallocation for Neural Machine Translation
Sugyeong Eo, Chanjun Park · Mathematics · 2026
Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.
Mathematics · Multidisciplinary Digital Publishing Institute · article · 2026
Abstract
The advent of large language models has reshaped the landscape of artificial intelligence, yet their learning dynamics remain constrained by rigid training strategies. Curriculum learning (CL), inspired by the human learning process, improves model performance over conventional randomly shuffled training while incurring no additional computational overhead. However, its competence-conservative mechanism often leads to diminished learning stimuli and suboptimal performance plateaus. Inspired by the flow theory in psychology, this study proposes in-batch hard sample injection curriculum learning (DACL), a learning strategy that dynamically balances stability and challenge.
DACL regulates sample selection by aligning the model’s competence with the intrinsic complexity of the data, allocating the Reasy proportion of each batch to instances within the competence range and the remaining (1−Reasy) to higher-difficulty samples that stimulate adaptive learning. Experiments on the English–Vietnamese pair demonstrate that DACL achieves superior performance over curriculum learning baselines across multiple difficulty evaluation criteria. Further experiments reveal the effectiveness of the similarity-based difficulty standard, demonstrating the ability to capture task complexity with greater precision.
Open the resolved locator · 10.3390/math14101607
Metadata licence: OpenAlex metadata (CC0)