Methods › General › Miscellaneous Components › Highway Layer

Highway Layer

108 papers tagged archive 2025-07-28

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

A Highway Layer contains an information highway to other layers that helps with information flow. It is characterised by the use of a gating unit to help this information flow.

A plain feedforward neural network typically consists of L layers where the $l$th layer (l ∈{1, 2, …, L}) applies a nonlinear transform H (parameterized by 𝐖_(𝐇,𝐥)) on its input 𝐱ₗ to produce its output 𝐲ₗ. Thus, 𝐱₁ is the input to the network and 𝐲_𝐋 is the network’s output. Omitting the layer index and biases for clarity,

𝐲 = H(𝐱,𝐖_𝐇)

H is usually an affine transform followed by a non-linear activation function, but in general it may take other forms.

For a highway network, we additionally define two nonlinear transforms T(𝐱,𝐖_𝐓) and C(𝐱,𝐖_𝐂) such that:

𝐲 = H(𝐱,𝐖_𝐇)·T(𝐱,𝐖_𝐓) + 𝐱·C(𝐱,𝐖_𝐂)

We refer to T as the transform gate and C as the carry gate, since they express how much of the output is produced by transforming the input and carrying it, respectively. In the original paper, the authors set C = 1 − T, giving:

𝐲 = H(𝐱,𝐖_𝐇)·T(𝐱,𝐖_𝐓) + 𝐱·(1-T(𝐱,𝐖_𝐓))

The authors set:

T(x) = σ(𝐖_𝐓ᵀ𝐱 + 𝐛_𝐓)

Image: Sik-Ho Tsang

Source: Highway NetworksSee Code · kefirski/pytorch_Highway

Papers archive 2025-07-28

30 shown of 108, 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 96 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
Speech Synthesis44
Text to Speech41
text-to-speech41
Text-To-Speech Synthesis15
Decoder11
Speech Recognition9
Language Modelling8
speech-recognition8
Language Modeling6
Sentence6
Transfer Learning5
Voice Cloning5
Data Augmentation4
Voice Conversion4
Audio Synthesis3
CPU3
Diversity3
Expressive Speech Synthesis3
GPU3
General Classification3

Usage over time archive 2025-07-28

Papers per year tagged with Highway Layer: 2015 to 2025, peak 21 21 0 2015: 3 papers 2015 2016: 1 paper 2016 2017: 6 papers 2017 2018: 15 papers 2018 2019: 11 papers 2019 2020: 21 papers 2020 2021: 14 papers 2021 2022: 10 papers 2022 2023: 15 papers 2023 2024: 11 papers 2024 2025: 1 paper 2025
Papers per year the archive tags with this method, by the paper's archive date (108 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

Miscellaneous Components

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