SOM-OLP introduces per-data-point continuous latent positions and a separable quadratic surrogate objective derived from STVQ, yielding efficient monotonic block coordinate descent with linear complexity and competitive performance on benchmarks.
SOM- V AE: Interpretable discrete representation learning on time series,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Self-Organizing Maps with Optimized Latent Positions
SOM-OLP introduces per-data-point continuous latent positions and a separable quadratic surrogate objective derived from STVQ, yielding efficient monotonic block coordinate descent with linear complexity and competitive performance on benchmarks.