# Low-Rank Compression for IMC Arrays

arXiv: https://arxiv.org/abs/2502.07820 · HTML: https://mlsanity.com/paper/2502.07820/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 19 references that are in the map.
- First posted on arXiv 10 Feb 2025.

## Builds on (most cited first, 7 of 7)

- [Learning both Weights and Connections for Efficient Neural Networks](https://mlsanity.com/paper/1506.02626/) (2015)
- [Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation](https://mlsanity.com/paper/1404.0736/) (2014)
- [PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning](https://mlsanity.com/paper/2001.00138/) (2020)
- [Any-Precision Deep Neural Networks](https://mlsanity.com/paper/1911.07346/) (2019)
- [DNN+NeuroSim V2.0: An End-to-End Benchmarking Framework for Compute-in-Memory Accelerators for On-chip Training](https://mlsanity.com/paper/2003.06471/) (2020)
- [TRP: Trained Rank Pruning for Efficient Deep Neural Networks](https://mlsanity.com/paper/2004.14566/) (2020)
- [Training CNNs with Low-Rank Filters for Efficient Image Classification](https://mlsanity.com/paper/1511.06744/) (2015)

## Built on by (most cited first, 2 of 2)

- [Row-Column Hybrid Grouping for Fault-Resilient Multi-Bit Weight Representation on IMC Arrays](https://mlsanity.com/paper/2508.15685/) (2025)
- [CanniUplift: A Holistic Framework for Mitigating Seller and Incentive Cannibalization in E-commerce Uplift Modeling](https://mlsanity.com/paper/2607.05242/) (2026)

## Abstract (arXiv)

In this study, we address the challenge of low-rank model compression in the context of in-memory computing (IMC) architectures. Traditional pruning approaches, while effective in model size reduction, necessitate additional peripheral circuitry to manage complex dataflows and mitigate dislocation issues, leading to increased area and energy overheads. To circumvent these drawbacks, we propose leveraging low-rank compression techniques, which, unlike pruning, streamline the dataflow and seamlessly integrate with IMC architectures. However, low-rank compression presents its own set of challenges, namely i) suboptimal IMC array utilization and ii) compromised accuracy. To address these issues, we introduce a novel approach i) employing shift and duplicate kernel (SDK) mapping technique, which exploits idle IMC columns for parallel processing, and ii) group low-rank convolution, which mitigates the information imbalance in the decomposed matrices. Our experimental results demonstrate that our proposed method achieves up to 2.5x speedup or +20.9% accuracy boost over existing pruning techniques.