Pith. sign in

REVIEW

Empirical Evaluation of Biased Methods for Alpha Divergence Minimization

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2105.06587 v1 pith:MKM64LCZ submitted 2021-05-13 cs.LG stat.ML

Empirical Evaluation of Biased Methods for Alpha Divergence Minimization

classification cs.LG stat.ML
keywords biasedmethodssolutionsalpha-divergencebiasminimizationactuallyaffects
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In this paper we empirically evaluate biased methods for alpha-divergence minimization. In particular, we focus on how the bias affects the final solutions found, and how this depends on the dimensionality of the problem. We find that (i) solutions returned by these methods appear to be strongly biased towards minimizers of the traditional "exclusive" KL-divergence, KL(q||p), and (ii) in high dimensions, an impractically large amount of computation is needed to mitigate this bias and obtain solutions that actually minimize the alpha-divergence of interest.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.