The authors designed learnable spectral activations (LSA) for implicit neural representations (INRs).
Learned activations reshape spectral tuning in neural fields
A residual Fourier series at each neuron allows independent control of frequency content without expanding the representable function class.
Big Tech
Tamir Shor · Or Litany · Alex Bronstein
Technion – Israel Institute of Technology · NVIDIA · Institute of Science and Technology Austria (ISTA)
Research Digest··3 min read
The authors introduce learnable spectral activations (LSA), which replace fixed nonlinearities in implicit neural representations (INRs) with a layer-wise truncated Fourier series whose harmonic amplitudes are learned alongside network weights.
Why this paper
From NVIDIA and 2 others
In one line
Learnable spectral activations improve implicit neural representation fitting by separating feature selection from spectral shaping.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§