Methods › General › Regularization › Manifold Mixup

Manifold Mixup

26 papers tagged archive 2025-07-28

Introduced by Vikas Verma et al. in Manifold Mixup: Better Representations by Interpolating Hidden States

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

Manifold Mixup is a regularization method that encourages neural networks to predict less confidently on interpolations of hidden representations. It leverages semantic interpolations as an additional training signal, obtaining neural networks with smoother decision boundaries at multiple levels of representation. As a result, neural networks trained with Manifold Mixup learn class-representations with fewer directions of variance.

Consider training a deep neural network f(x) = fₖ(gₖ(x)), where gₖ denotes the part of the neural network mapping the input data to the hidden representation at layer k, and fₖ denotes the part mapping such hidden representation to the output f(x). Training f using Manifold Mixup is performed in five steps:

(1) Select a random layer k from a set of eligible layers S in the neural network. This set may include the input layer g₀(x).

(2) Process two random data minibatches (x, y) and (x′, y′) as usual, until reaching layer k. This provides us with two intermediate minibatches (gₖ(x), y) and (gₖ(x′), y′).

(3) Perform Input Mixup on these intermediate minibatches. This produces the mixed minibatch:

(g̃ₖ, ỹ) = (Mix_λ(gₖ(x), gₖ(x′)), Mix_λ(y, y′)),

where Mix_λ(a, b) = λ·a + (1 − λ) ·b. Here, (y, y′ ) are one-hot labels, and the mixing coefficient λ∼Beta(α, α) as in mixup. For instance, α= 1.0 is equivalent to sampling λ∼U(0, 1).

(4) Continue the forward pass in the network from layer k until the output using the mixed minibatch (g̃ₖ, ỹ).

(5) This output is used to compute the loss value and gradients that update all the parameters of the neural network.

PaperSourceSee Code · vikasverma1077/manifold_mixup

Papers archive 2025-07-28

26 shown of 26, 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.

Tasks archive 2025-07-28

20 shown of 57 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Data Augmentation7
Image Classification6
image-classification5
General Classification3
Few-Shot Learning2
MUlTI-LABEL-ClASSIFICATION2
Meta-Learning2
Multi-Label Classification2
Representation Learning2
Text Classification2
text-classification2
3D Point Cloud Classification1
3D Point Cloud Data Augmentation1
Active Learning1
Anomaly Detection1
Benchmarking1
Bilevel Optimization1
Classification1
Code Classification1
Computational Efficiency1

Usage over time archive 2025-07-28

Papers per year tagged with Manifold Mixup: 2018 to 2025, peak 5 5 0 2018: 1 paper 2018 2019: 3 papers 2019 2020: 5 papers 2020 2021: 5 papers 2021 2022: 5 papers 2022 2023: 2 papers 2023 2024: 4 papers 2024 2025: 1 paper 2025
Papers per year the archive tags with this method, by the paper's archive date (26 dated). Bars are counts, not a trend claim.

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

Regularization

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