# Generalized Max Pooling

arXiv: https://arxiv.org/abs/1406.0312 · HTML: https://mlsanity.com/paper/1406.0312/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Jun 2014.

## Built on by (most cited first, 12 of 12)

- [Keep It SimPool: Who Said Supervised Transformers Suffer from Attention Deficit?](https://mlsanity.com/paper/2309.06891/) (2023)
- [Exploiting Low-Rank Tensor-Train Deep Neural Networks Based on Riemannian Gradient Descent With Illustrations of Speech Processing](https://mlsanity.com/paper/2203.06031/) (2022)
- [CA-I2P: Channel-Adaptive Registration Network with Global Optimal Selection](https://mlsanity.com/paper/2506.21364/) (2025)
- [Self-Supervised Vision Transformers for Writer Retrieval](https://mlsanity.com/paper/2409.00751/) (2024)
- [KaiRacters: Character-level-based Writer Retrieval for Greek Papyri](https://mlsanity.com/paper/2407.07536/) (2024)
- [Set-Based Face Recognition Beyond Disentanglement: Burstiness Suppression With Variance Vocabulary](https://mlsanity.com/paper/2304.06249/) (2023)
- [VCR: Variance-Driven Channel Recalibration for Robust Low-Light Enhancement](https://mlsanity.com/paper/2603.10975/) (2026)
- [Deformable-Heatmap-Segmentation for Automobile Visual Perception](https://mlsanity.com/paper/2407.07493/) (2024)
- [Shrinkage Initialization for Smooth Learning of Neural Networks](https://mlsanity.com/paper/2504.09107/) (2025)
- [On the Burstiness of Faces in Set](https://mlsanity.com/paper/2506.20312/) (2025)
- [EMPEROR: Efficient Moment-Preserving Representation of Distributions](https://mlsanity.com/paper/2509.16379/) (2025)
- [ICDAR 2026 Competition on Writer Identification and Pen Classification from Hand-Drawn Circles](https://mlsanity.com/paper/2605.07816/) (2026)

## Abstract (arXiv)

State-of-the-art patch-based image representations involve a pooling operation that aggregates statistics computed from local descriptors. Standard pooling operations include sum- and max-pooling. Sum-pooling lacks discriminability because the resulting representation is strongly influenced by frequent yet often uninformative descriptors, but only weakly influenced by rare yet potentially highly-informative ones. Max-pooling equalizes the influence of frequent and rare descriptors but is only applicable to representations that rely on count statistics, such as the bag-of-visual-words (BOV) and its soft- and sparse-coding extensions. We propose a novel pooling mechanism that achieves the same effect as max-pooling but is applicable beyond the BOV and especially to the state-of-the-art Fisher Vector -- hence the name Generalized Max Pooling (GMP). It involves equalizing the similarity between each patch and the pooled representation, which is shown to be equivalent to re-weighting the per-patch statistics. We show on five public image classification benchmarks that the proposed GMP can lead to significant performance gains with respect to heuristic alternatives.