- Rating
- 1
- Confidence
- 2
- Informativeness
- 1
- Presentation
- 1
- Soundness
- 1
Read review
Paper summary
The authors build a normalizing flow from Laplace KDEs. The main benefit here is the closed-form inverse obtained by solving a quadratic within each interval between 'knots' (as they call them). They evaluate both data-only and trained versions.
Critical strengths
- the closed-form inverse is useful: it avoids the numerical inversion required by the Gaussianization-flow. (though i'm not sure how novel this is?)
Critical weaknesses
- the paper is difficult to read, which makes it difficult to assess
- some central theoretical claims are dubious as stated. One proposition claims that taking both bandwidths to zero causes a diverging derivative ratio. But if the two mixtures are identical, then the map T is the identity, which implies that ratio is always 1.
- fastICA is still stronger
Questions for authors
- what assumptions are missing from Propositions 2 and 4?
- What supports the claim that “two quantization paths beat one”? In Table 2, freezing h_2 gives better results than learning both bandwidths at r=10, r=100, and HEPMASS k=5
2 comments
Responses to the Reviewer - part 2
Q2. What supports "two quantization paths beat one"?
Nothing in Table 2 supports this sentence; it also contradicts our own App. F.1 ("the second mixture buys nothing here"), and we withdraw it. What Table 2 does support is that a learnable bandwidth beats none: pinning every bandwidth is far worse on the sharp targets (excess 0.585 vs 0.017 at ; test NLL vs at ). Two aspects of this comparison require clarification: (i) the " frozen" arm is the link, which at matched parameter count gets twice the inner knots ( vs 16; 128 vs 64 on HEPMASS) and a fixed outer map, so Table 2 does not isolate the number of quantization paths; (ii) on average the arm is ahead on the 1-D atoms at every (in all three seeds at and 30, in two of three at ), but by only 0.002–0.006 nats against a 0.09–0.57 gap to "both frozen"; on HEPMASS the sign alternates with (two paths better at , worse at ); and on 4-bit MNIST patches, a naturally discrete target, the two-path model is better by nats in 3/3 seeds (94.09 vs 95.92). The revised Sec. 6.3 says that a learnable bandwidth is what wins and that, between the two arms, the sign depends on the target. Prop. 4's "two independent directions" is a structural statement (each path alone reaches a quantizer). We also corrected Remark 1: a rational-quadratic spline does reach a staircase, as its knot derivatives diverge, so the difference is one of parameterization (a single temperature per path, levels set by the estimated laws), not of function class. To test the two paths empirically we ran a factorial at identical architecture ( and each learned or frozen at , same and parameter count, 5 seeds, all Table 2 targets and 4-bit patches; new table in App. F.1). The answer depends on the target. On the sharp 1-D atoms () the domain path alone matches both learned (on average within 0.005 nats), while the range path alone trails by 0.04 and 0.18 nats. On HEPMASS with five restored columns and on 4-bit patches both paths are needed at the same : two learned paths beat each single one in all 5 seeds (by 1.56 and 8.93 nats at ; by 1.15 and 0.33 nats on 4-bit patches, where the range path alone recovers 85% of the gain); with twice the inner knots the link is again ahead at (0.57 nats, 5/5 seeds). At and on standard HEPMASS, with little to quantize, the four factorial arms are within 0.11 nats. The revision reports this per target, states the second path as a property of the class rather than a general gain, and softens the abstract accordingly.
Readability.
We agree. The revision compresses Secs. 3–6 and cuts part of the process narration from the main text; Sec. 6 keeps its opening roadmap from claims to experiments. For the final version we will add a notation table (knots, weights, bandwidths, levels , , cells) and one running example carried through Secs. 3–4.
Some theoretical claims are dubious as stated. See Q1 above: the statements are corrected, the missing hypotheses are stated, and the corrected statements are checked numerically.
FastICA is stronger.
Agreed, and the paper says so (Sec. 6.4: "FastICA is strongest overall", Tab. 3 caption). Our claim is the narrower within-search comparison: inside RADICAL's exhaustive Jacobi search, kernel marginals beat the m-spacing entropy estimate in 13 of 20 cells and on skewed sources at every dimension (App. G). The contributions paragraph and Sec. 6.4 of the revision state this scope explicitly.
Novelty of the closed-form inverse. Rational-quadratic splines also invert analytically; among mixture-CDF layers (GF, Flow++), which invert by bisection, the Laplace kernel is what makes the inverse closed-form and equal-cost (Sec. 3; Lem. 2, Thm. 1).
Responses to the Reviewer - part 1
Q1. Missing assumptions in Props. 2 and 4.
You are right, and thank you for the counterexample: as stated, Prop. 4(iii) is false when the two mixtures coincide, and the "not C2" part of Prop. 2 fails in the same degenerate cases. The missing hypotheses are:
We verified the corrected statements numerically in 400-digit arithmetic (for small h the effect is below float64 resolution), with a script added to the supplementary code: identical and translated pairs give log10(supT′/infT′)=0.00 at all seven tested h from 1 to 0.01; four non-translate pairs reach 13.0–51.8 at h=0.01, with fitted rates c=0.300, 1.200, 0.677 and 0.993 (the first two where the case analysis predicts 0.3 and 1.2); T′′ jumps vanish exactly for identical and affine pairs and are nonzero at every inner knot of a generic pair; off the breakpoints dist(T(x),{μ2})/h2≤7.51 for h2 from 0.1 to 0.003, whereas at the breakpoints it grows from 3.9 to 133. Auditing the remaining statements, we also make explicit that Prop. 6 assumes uniform weights 1/Ki; Def. 1 now requires wk>0, which Prop. 5 also needs.