Methods › General › Attention Mechanisms › BAM
Bottleneck Attention Module
BAM
Introduced by Jongchan Park et al. in BAM: Bottleneck Attention Module
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
Park et al. proposed the bottleneck attention module (BAM), aiming to efficiently improve the representational capability of networks. It uses dilated convolution to enlarge the receptive field of the spatial attention sub-module, and build a bottleneck structure as suggested by ResNet to save computational cost.
For a given input feature map X, BAM infers the channel attention s_c ∈ℝ^C and spatial attention sₛ∈ℝ^(H×W) in two parallel streams, then sums the two attention maps after resizing both branch outputs to ℝ^(C×H ×W). The channel attention branch, like an SE block, applies global average pooling to the feature map to aggregate global information, and then uses an MLP with channel dimensionality reduction. In order to utilize contextual information effectively, the spatial attention branch combines a bottleneck structure and dilated convolutions. Overall, BAM can be written as s_c = BN(W₂(W₁GAP(X)+b₁)+b₂)
sₛ = BN(Conv₂^(1 ×1)(DC₂^(3×3)(DC₁^(3 ×3)(Conv₁^(1 ×1)(X))))) s = σ(Expand(sₛ)+Expand(s_c)) Y = s X+X where Wᵢ, bᵢ denote weights and biases of fully connected layers respectively, Conv₁^(1×1) and Conv₂^(1×1) are convolution layers used for channel reduction. DCᵢ^(3×3) denotes a dilated convolution with 3×3 kernel, applied to utilize contextual information effectively. Expand expands the attention maps sₛ and s_c to ℝ^(C×H×W).
BAM can emphasize or suppress features in both spatial and channel dimensions, as well as improving the representational power. Dimensional reduction applied to both channel and spatial attention branches enables it to be integrated with any convolutional neural network with little extra computational cost. However, although dilated convolutions enlarge the receptive field effectively, it still fails to capture long-range contextual information as well as encoding cross-domain relationships.
Papers archive 2025-07-28
30 shown of 33, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.
-
Synthesizing Images on Perceptual Boundaries of ANNs for Uncovering and Manipulating Human Perceptual Variability 6 May 2025 · 0 repositories · arXiv:2505.03641
-
Reliability Assessment of Low-Cost PM Sensors under High Humidity and High PM Level Outdoor Conditions 9 Apr 2025 · 0 repositories · arXiv:2504.06615
-
AdaCS: Adaptive Normalization for Enhanced Code-Switching ASR 13 Jan 2025 · 1 repository · arXiv:2501.07102
-
Scale-wise Bidirectional Alignment Network for Referring Remote Sensing Image Segmentation 1 Jan 2025 · 0 repositories · arXiv:2501.00851
-
Greenback Bears and Fiscal Hawks: Finance is a Jungle and Text Embeddings Must Adapt 11 Nov 2024 · 0 repositories · arXiv:2411.07142
-
Batch, match, and patch: low-rank approximations for score-based variational inference 29 Oct 2024 · 1 repository · arXiv:2410.22292
-
BAM! Just Like That: Simple and Efficient Parameter Upcycling for Mixture of Experts 15 Aug 2024 · 0 repositories · arXiv:2408.08274
-
Unsupervised Representation Learning by Balanced Self Attention Matching 4 Aug 2024 · 1 repository · arXiv:2408.02014
-
Mitigating Catastrophic Forgetting in Language Transfer via Model Merging 11 Jul 2024 · 0 repositories · arXiv:2407.08699
-
BAM: Box Abstraction Monitors for Real-time OoD Detection in Object Detection 27 Mar 2024 · 0 repositories · arXiv:2403.18373
-
Methylation Operation Wizard (MeOW): Identification of differentially methylated regions in long-read sequencing data 27 Feb 2024 · 0 repositories · arXiv:2402.17182
-
Batch and match: black-box variational inference with a score-based divergence 22 Feb 2024 · 2 repositories · arXiv:2402.14758Syntology ran 7 of 10 samples · 3 unverified
-
MS-Former: Memory-Supported Transformer for Weakly Supervised Change Detection with Patch-Level Annotations 16 Nov 2023 · 1 repository · arXiv:2311.09726
-
Bias Amplification Enhances Minority Group Performance 13 Sep 2023 · 1 repository · arXiv:2309.06717Syntology ran 7 of 9 samples · 2 unverified · 9 pointer-only (licence)
-
Bias-Aware Minimisation: Understanding and Mitigating Estimator Bias in Private SGD 23 Aug 2023 · 0 repositories · arXiv:2308.12018
-
Boundary Attention Mapping (BAM): Fine-grained saliency maps for segmentation of Burn Injuries 24 May 2023 · 0 repositories · arXiv:2305.15365
-
A Convolutional-Transformer Network for Crack Segmentation with Boundary Awareness 23 Feb 2023 · 1 repository · arXiv:2302.11728
-
Interpretable Diabetic Retinopathy Diagnosis based on Biomarker Activation Map 13 Dec 2022 · 0 repositories · arXiv:2212.06299
-
Reinforcement Learning Agent Design and Optimization with Bandwidth Allocation Model 23 Nov 2022 · 0 repositories · arXiv:2211.12987
-
Thermodynamics of bidirectional associative memories 17 Nov 2022 · 0 repositories · arXiv:2211.09694
-
Identifying OCRs in cfDNA WGS Data by Correlation Clustering 19 Feb 2022 · 0 repositories · arXiv:2202.09618
-
BAM: Bayes with Adaptive Memory 4 Feb 2022 · 0 repositories · arXiv:2202.02405
-
BAM: Bayes Augmented with Memory 29 Sep 2021 · 0 repositories
-
Few-Shot Fine-Grained Action Recognition via Bidirectional Attention and Contrastive Meta-Learning 15 Aug 2021 · 1 repository · arXiv:2108.06647
-
BAM: A Balanced Attention Mechanism for Single Image Super Resolution 15 Apr 2021 · 1 repository · arXiv:2104.07566
-
LRez: C++ API and toolkit for analyzing and managing Linked-Reads data 26 Mar 2021 · 1 repository · arXiv:2103.14419
-
Blind Mask to Improve Intelligibility of Non-Stationary Noisy Speech 20 Aug 2020 · 0 repositories · arXiv:2008.09175
-
Bilateral Attention Network for RGB-D Salient Object Detection 30 Apr 2020 · 1 repository · arXiv:2004.14582
-
Benchmarking Attribution Methods with Relative Feature Importance 23 Jul 2019 · 2 repositories · arXiv:1907.09701Syntology ran 1 of 6 samples · 5 unverified
-
Interactive Learning of Environment Dynamics for Sequential Tasks 19 Jul 2019 · 0 repositories · arXiv:1907.08478
Tasks archive 2025-07-28
20 shown of 55 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections