- Rating
- 3
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes SurF, a model for forecasting irregularly sampled event streams. It is built on the Time Rescaling Theorem (TRT), which maps every point process to the same i.i.d. unit-rate exponential target. SurF parameterizes the cumulative intensity directly and uses a shared Transformer encoder to learn this transformation across multiple event-stream datasets. It introduces three variants: SurF-MoE and SurF-CSB, which provide closed-form cumulative intensities, and SurF-GLQ, which uses deterministic Gauss–Legendre quadrature. This design avoids the stochastic Monte Carlo integration commonly used by neural temporal point processes (TPPs). Experiments on six real-world datasets evaluate next-event prediction, multi-step forecasting, and leave-one-dataset-out transfer. The results are promising: SurF performs competitively with specialized TPP models and achieves strong zero-shot event-time forecasting results when the target dataset is excluded from training.
Critical strengths
- The idea is quite interesting and reasonable to use the Time Rescaling Theorem (TRT), which transforms event inter-arrival times into a common unit-rate exponential distribution when the conditional intensity is correctly specified.
- This paper explores whether a shared model can learn temporal patterns that transfer across datasets with different time scales and event dynamics. The leave-one-out evaluation is particularly useful.
- The experiments are convincing, and the evaluations cover different aspects.
Critical weaknesses
- More analysis should be provided to discuss the transferability of datasets. This could reveal whether cross-dataset gains arise from broadly shared dynamics or from a small number of particularly helpful source datasets.
- The intensity-floor assumption may restrict applicability to some real-world processes. The paper's practical claims would be more convincing if the model were evaluated on datasets or controlled synthetic processes with long inactive intervals.
- The efficiency comparison should make clear which costs come from the cumulative-intensity parameterization, the Transformer encoder, and the numerical inversion process.
Questions for authors
- Can the authors provide evidence separating the effect of the shared exponential target from that of the shared Transformer representation?
- Are the proposed masking or learnable-floor schemes evaluated on processes with long inactive intervals?
3 comments
W3: "The efficiency comparison should make clear which costs come from the cumulative-intensity parameterization, the Transformer encoder, and the numerical inversion process."
We profiled each component's share of time on Amazon with the paper's configurations.
| Share of time | MoE | CSB | GLQ |
|---|---|---|---|
| Train: encoder forward | 8% | 9% | 5% |
| Train: Λ head forward | 2% | 2% | 7% |
| Train: backward | 63% | 63% | 60% |
| Inference: encoder | 42% | 42% | 34% |
| Inference: Λ head | 6% | 6% | <2% |
| Inference: inversion | 0.5% | 0.5% | 0.4% |
- The closed-form parameterization is almost free. The MoE and CSB heads are a single linear layer followed by closed-form operations (App. I.2); they take 2% of a training step and 6% of inference, and inversion takes 0.5% of inference. Of the three components, the encoder is the largest in both training (8–9%) and inference (42%). The parameterization therefore adds little to SurF-MoE's cost, which trains 10.9–12.4× faster than NJDTPP and about 1.4× faster than DTPP (Table 4).
- GLQ's extra cost is quadrature: its head forward is 7% of a training step, and backpropagation runs through all Q nodes, a cost the closed-form variants avoid. At inference, its head and inversion together still take under 2.5%.
- The remaining 26–28% of a training step, and 51.5% (MoE, CSB) and over 63% (GLQ) of an inference pass, is spent outside the encoder, Λ head and inversion.
Revision to the paper: We will add this breakdown and use it to replace the Newton-time estimate in App. E.1.
Q1: "Can the authors provide evidence separating the effect of the shared exponential target from that of the shared Transformer representation?"
In SurF, the Exp(1) target is not a separate loss term but the likelihood itself (App. B), so it cannot be switched off on its own. Instead, we varied one factor at a time with the others held fixed:
| Factor varied | Held fixed | Change | Result |
|---|---|---|---|
| Target | one corpus at a time (no pooling), encoder, constant-hazard head | MSE regression → exponential likelihood | RMSE lower on 4/6 corpora in every seed: Taxi −15%, Amazon −12%, Retweet −3%, Earthquake −3% |
| Representation (W1) | CSB head, likelihood, target corpus held out | 1 → 5 training corpora | vs median single source: held-out RMSE −4% to −62% on 5/6 targets; NLL lower on 6/6 |
| Head (RU4Q, W2) | target, encoder, multi-corpus training | FullyNN- / IFTPP-style head → SurF head | RMSE −17% to −76%, lower on 6/6 corpora for every SurF head |
- Target: changing only the target, with no pooling, lowers RMSE on 4/6 corpora in every seed, even though MSE trains directly on what RMSE measures.
- Representation: changing only the pooling, with the target fixed, lowers held-out RMSE on 5/6 targets and NLL on 6/6.
- Complementary: both factors help, and the unit-free target is what makes it possible to pool corpora whose time units range from seconds to days (Sec. 3.2).
- Head: with both held fixed, swapping in a SurF head adds a third gain, lowering RMSE by 17–76% on all six corpora.
Revision to the paper: We will add these ablations to Sec. 4.4.
Thanks for the detailed additional experiments, which address most of my concerns.
We thank the reviewer for finding the idea interesting, the leave-one-out evaluation useful, and the experiments convincing. We address each concern below with new experiments.
W1: "More analysis should be provided to discuss the transferability of datasets. This could reveal whether cross-dataset gains arise from broadly shared dynamics or from a small number of particularly helpful source datasets."
They come from dynamics shared across the pool, not from one or two sources. We tested this with the CSB head, measuring held-out RMSE (deterministic rollout at z = 1) after training on each corpus alone (3 seeds) and, for every target, after removing each source corpus in turn (30 runs).
Revision to the paper: We will add the full single-source and leave-one-source-out analysis to the appendix.
W2 / Q2: "The intensity-floor assumption may restrict applicability to some real-world processes. [...] learnable-floor schemes evaluated on processes with long inactive intervals?"
This is a fair concern. The floor is part of the model, not an assumption about the data, so it does not require the data to have no inactive periods. It makes Λθ a bijection that can be inverted for sampling (Theorem 2, applied to the model); without it, MoE's Λ is bounded and cannot be inverted for large z (Remark, App. A). Its cost is also bounded: App. A.1 bounds its likelihood bias by T⋅λfloor, so a floor of 10−4 represents long zero-intensity stretches at a cost of at most 10−4 nats per unit time.
To test this empirically, we compared no, fixed (10⁻⁴) and learnable floors on four processes with exact oracle likelihoods:
The table shows excess NLL over the oracle, in nats (mean of 2 seeds; GLQ scored with the exact integral).
Revision to the paper: We will add this study, including the learnable floor, to App. A.1.
Continued in the next comment ....