acceptodds
Withdrawn from ICLR 2027

Preserving Decision-Relevant History: Response-Rank Memory for Write-Before-Query Prediction

Withdrawn from ICLR 2027, so left out of the paper lists and the map.

Abstract

Before a query is known, a fixed-size writer must preserve the history distinctions that determine a later response. We formulate this write-before-query interface as compression of a history-indexed response field, and derive its linear response-operator geometry, capacity–allocation decomposition, and late-query timing gap. Query-Rank Memory (QRM) is the corresponding spectral writer with nested prefixes, estimated by weighted reduced-rank regression. Across six native history conditions, QRM lowers response MSE in five, while controlled, exchange, and known-effect tests connect the same response field to storage choices and decisions.

Primary area:
learning theory
Keywords:
write-before-query memory, response operator, memory compression, predictive state representation, reduced-rank regression, representation learning

Reviews

Among:
Shape:
1234this paper, Bottom 1%
Figure 1. Mean review rating of the reviewed papers at ICLR 2027. This paper averages 1.00 over 3 reviews, above 0% of the others.
Reviewer_R56MOpenReview ↗
Rating
1
Confidence
3
Informativeness
1
Presentation
2
Soundness
1
Read review
Paper summary

The paper studies a genuinely interesting interface: a fixed-size writer must commit a state before it knows which query will be asked, so the state should preserve the history distinctions that change the answer, not the history variance. The formal object is a history-indexed response field Φ(H), and the central identity is that the width-k response error splits into an unavoidable spectral tail plus an allocation term, L(P) = tr((I−P)C) = Σ_{j>k} λ_j(C) + [Σ_{j≤k} λ_j(C) − tr(PC)]. That is correct, and the paper makes it measurable: a fixed-width exchange identity, a training-side score for proposed swaps, a Clopper–Pearson width certificate, and a 342-row frozen results table. The protocol discipline is well above the norm for this kind of paper — a pre-declared exchange rule, random-replacement controls, an explicit negative control for zero-overlap query families, and a disclosed failure (PH Lift) rather than a hidden one.

Critical strengths

S1: The decomposition is correct, and the paper demonstrates it with a construction rather than an assertion. S2: Protocol discipline that most papers in this area do not have. S3: The calibration mathematics is exactly right. S4: Two results that damage the paper's own case are reported rather than buried. S5: The dependence experiment is the right diagnostic.

Critical weaknesses

The difficulty is that the method is weighted multiresponse reduced-rank regression, which the paper says explicitly ("the estimator is weighted multiresponse reduced-rank regression"; "Under matched features, response weights, and a linear reader, QRM recovers the corresponding population subspace"), and the theory is the Eckart–Young/Ky Fan variational identity plus a conditional-expectation argument. In the one controlled experiment where the response directions are known, QRM and the raw-coordinate coefficient SVD agree to every printed digit (186.1271 / 2.33×10⁻⁵ / 0.4966 / 0.5000), and the only component that distinguishes QRM from that baseline — family-equal column weighting — has all three ablation intervals containing zero. What remains as evidence is a comparison against PCA on six robomimic conditions, and in that comparison a plain GRU with a 32-byte state beats QRM at 64 bytes on all six.

Three further results inside the paper work against the framing rather than for it. The neural controls show full history losing to current-frame-only in 8 of 10 comparisons, which is the opposite of the paper's premise. The width sweep is non-monotone — on Lift, QRM's response error rises from 0.5701 at width 4 to 1.0388 at width 32, 1.11× worse than storing nothing — which cannot happen for a nested prefix family under Theorem 2 and points at reader-fitting instability in a single-seed pipeline. And the calibration certificate, presented as an under-powered sample ("no width is selected"), inverts to 49–55 of 60 episodes failing in one query family, i.e. the 0.05 tolerance is below every method's achieved error and no amount of episodes would certify it.

Questions for authors
  1. Table 2 reports a 32 B GRU beating the 64 B QRM state on all six conditions. What does QRM offer a practitioner who can afford a learned encoder at the same state size — and if the answer is "interpretability" or "no training beyond a linear fit", say that in the abstract.
  2. Why does QRM's response error rise from 0.5701 at width 4 to 1.0388 at width 32 on recorded Lift, when the prefixes are nested and tr((I−P)C) is non-increasing? Is per-width ridge selection responsible, and what width do you actually recommend for that task?
  3. Table 11's minimum joint bound of 0.983274 corresponds to 55 of 60 episodes failing, and the minimum scalar bound of 0.152704 to 2 of 60. Which query family fails, and why is the outcome attributed to the 55-episode requirement rather than to a 0.05 tolerance that no method meets?
  4. In the 564-row intervention grid, 47 of 504 targeted/random rows are positive. What is the positive rate for targeted swaps alone, and under what conditions does Proposition 5's ideal gain survive the fitted-reader term in Eq. (19)?
  5. Given Table 3, what is the method's scope? If the deployed query policy conditions on the history — which is what "retrieve the blue object" implies in a closed-loop agent — the proposed allocation is 2.16× worse than a joint-loss search. Do you claim the independent-query regime only?
  6. Table 13 shows current-only beating full-history in 8 of 10 neural comparisons. How do you reconcile that with the paper's premise, and does the linear current-frame comparison (0.0149, 0.0166) measure the same thing?
  7. What is the seed-to-seed spread of the QRM−PCA gap? Please report it per seed for the six conditions, since the entire claimed effect on PH Can is 0.0051.
  8. Table 15's three intervals all contain zero, and Table 1 shows QRM identical to coefficient SVD. What evidence remains that response-weighted allocation, as opposed to whitening plus reduced-rank regression, is doing the work?
  9. Can the certificate be made informative by lowering its ambition — for example, certifying "no worse than the no-memory floor" instead of "below 0.05" — and would that change any of the six external outcomes?
Reviewer_edSxOpenReview ↗
Rating
1
Confidence
2
Informativeness
1
Presentation
1
Soundness
1
Read review
Paper summary

The paper studies the problem of compressing history into a small number of states without knowing what the downstream task will be. They propose (?) QRM as a means to do this.

Critical strengths

None.

Critical weaknesses

The paper is extremely poorly written, the text reads essentially like the unfiltered output of an LLM (and not a particularly good one at that). There is no attempt to state a clear problem formulation or motivate and situate the problem within the context of prior work. This is unfortunately so detrimental to the experience of reading the paper that I am unable to judge any of the possible merits it has.

Questions for authors

Unless the authors make any kind of effort to define and state their problem, goals and results in human-understandable language, I do not think any specific questions could change my assessment.

Reviewer_jGEYOpenReview ↗
Rating
1
Confidence
4
Informativeness
1
Presentation
1
Soundness
1
Read review
Paper summary

The abstract and the introduction both use language and terminology I am not very familiar with and assume a knowledge of a problem that the authors do not themselves provide.

I did spend some time looking at the empirical results. In Table 1 the authors report a "Bayes Risk" that is never defined. Further, they report a "Target MSE" that is lower than "Bayes Risk".

In general, I do not think the authors put in enough effort in presenting their work to warrant any further review.

Critical strengths

I do not support acceptance.

Critical weaknesses

The presentation of the problem and empirical results are both below the standard required / expected by the ICLR community.

Questions for authors

I do not have any questions for the authors.

From OpenReview, CC BY 4.0. Authors as signed there; shown unedited.

Related papers

Discussion (0)

Sign in to comment.