# VCMM: Variance-Calibrated Momentum for Multimodal Learning

arXiv: https://arxiv.org/abs/2609.27577 · HTML: https://mlsanity.com/paper/2609.27577/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 33 references that are in the map.
- First posted on arXiv 23 Sep 2026.

## Builds on (most cited first, 25 of 25)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [Multimodal Machine Learning: A Survey and Taxonomy](https://mlsanity.com/paper/1705.09406/) (2017)
- [Look, Listen and Learn](https://mlsanity.com/paper/1705.08168/) (2017)
- [What Makes Training Multi-Modal Classification Networks Hard?](https://mlsanity.com/paper/1905.12681/) (2019)
- [Balanced Multimodal Learning via On-the-fly Gradient Modulation](https://mlsanity.com/paper/2203.15332/) (2022)
- [Multimodal Deep Learning](https://mlsanity.com/paper/2301.04856/) (2023)
- [Modality Competition: What Makes Joint Training of Multi-modal Network Fail in Deep Learning? (Provably)](https://mlsanity.com/paper/2203.12221/) (2022)
- [PMR: Prototypical Modal Rebalance for Multimodal Learning](https://mlsanity.com/paper/2211.07089/) (2022)
- [Characterizing and overcoming the greedy nature of learning in multi-modal deep neural networks](https://mlsanity.com/paper/2202.05306/) (2022)
- [Multimodal Representation Learning by Alternating Unimodal Adaptation](https://mlsanity.com/paper/2311.10707/) (2023)
- [Boosting Multi-modal Model Performance with Adaptive Gradient Modulation](https://mlsanity.com/paper/2308.07686/) (2023)
- [MMPareto: Boosting Multimodal Learning with Innocent Unimodal Assistance](https://mlsanity.com/paper/2405.17730/) (2024)
- [ReconBoost: Boosting Can Achieve Modality Reconcilement](https://mlsanity.com/paper/2405.09321/) (2024)
- [Understanding the Role of Momentum in Stochastic Gradient Methods](https://mlsanity.com/paper/1910.13962/) (2019)
- [Classifier-guided Gradient Modulation for Enhanced Multimodal Learning](https://mlsanity.com/paper/2411.01409/) (2024)
- [DecAlign: Hierarchical Cross-Modal Alignment for Decoupled Multimodal Representation Learning](https://mlsanity.com/paper/2503.11892/) (2025)
- [Adaptive Unimodal Regulation for Balanced Multimodal Information Acquisition](https://mlsanity.com/paper/2503.18595/) (2025)
- [YellowFin and the Art of Momentum Tuning](https://mlsanity.com/paper/1706.03471/) (2017)
- [Rethinking Multimodal Learning from the Perspective of Mitigating Classification Ability Disproportion](https://mlsanity.com/paper/2502.20120/) (2025)
- [Kalman Gradient Descent: Adaptive Variance Reduction in Stochastic Optimization](https://mlsanity.com/paper/1810.12273/) (2018)
- [Balance-aware Sequence Sampling Makes Multi-modal Learning Better](https://mlsanity.com/paper/2501.01470/) (2025)
- [Dynamic Momentum Recalibration in Online Gradient Learning](https://mlsanity.com/paper/2603.06120/) (2026)

## Abstract (arXiv)

Multimodal joint training often suffers from modality imbalance, where a dominant modality suppresses the optimization of others. Existing methods mainly balance modality learning by modulating gradient magnitudes or directions, modifying optimization objectives, or adjusting training strategies, with most interventions focusing on the current update. However, when combined with widely used momentum-based optimizers, the update also incorporates accumulated information from previous gradients, which is not explicitly addressed by current-step modulation alone. To address this issue, we propose Variance-Calibrated MomentuM (VCMM), which adapts gradient memory to modality-specific gradient dynamics. Specifically, VCMM estimates minibatch noise and temporal drift online and uses their relative strength to determine modality-specific momentum through a Kalman-inspired controller. We further center the control signal across modalities and apply exact bias correction for the time-varying first moment, enabling adaptive gradient memory without extra network passes or explicit learning-rate scaling. Experiments on four multimodal benchmarks demonstrate consistent improvements with modest training overhead.