arXiv Artificial Intelligence

Sample-Conditioned Representation Selection for Audio Few-Shot Learning

1 hour ago 16.09.2026 07:00 👁 1 views
Sample-Conditioned Representation Selection for Audio Few-Shot Learning

Quick summary

arXiv:2609.17076v1 Announce Type: new Abstract: Few-shot audio classifiers may rely on foreground-background co-occurrences and fail when those correlations shift. On SpurAudio, the resulting representation shift is concentrated and class dependent: for ResNet12, the top 10 percent of channels explain 82.80 percent of the null-corrected shift contribution. We propose SAMPLESELECT, which predicts a fixed-budget feature mask independently for each input while keeping the encoder and source classifier frozen. Training uses differentiable Gumbel Top-k selection with foreground classification and c

Key takeaways

  • arXiv:2609.17076v1 Announce Type: new Abstract: Few-shot audio classifiers may rely on foreground-background co-occurrences and fail when those correlations shift.
  • On SpurAudio, the resulting representation shift is concentrated and class dependent: for ResNet12, the top 10 percent of channels explain 82.80 percent of the null-corrected shift contribution.
  • We propose SAMPLESELECT, which predicts a fixed-budget feature mask independently for each input while keeping the encoder and source classifier frozen.

Why it matters

This development shows AI moving deeper into everyday software. Productivity potential should be weighed against price, data permissions, exportability and the preservation of human control.

Kaynak sitede devamını oku: arXiv Artificial Intelligence ↗