File size: 2,299 Bytes
d055dde
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
{
  "schema_version": 1,
  "title": "Latent Laplace Diffusion for Irregular Multivariate Time Series (LLapDiff)",
  "emoji": "🌀",
  "space_id": "snaykey/repro-llapdiff",
  "paper": {
    "arxiv_id": "2605.19805",
    "openreview_id": "t73XUJvyQr"
  },
  "tags": [
    "icml2026-repro",
    "paper-t73XUJvyQr"
  ],
  "updated_at": "2026-07-20T08:30:00+00:00",
  "root": {
    "slug": "index",
    "title": "Latent Laplace Diffusion for Irregular Multivariate Time Series (LLapDiff)",
    "file": "pages/index.md",
    "children": [
      {
        "slug": "claim-1-latent-trajectories",
        "title": "LLapDiff models irregular targets as low-dimensional latent trajectories, enabling horizon-wide generation without numerical integration over physical time (Section 4.4).",
        "file": "pages/claim-1-latent-trajectories/page.md",
        "children": []
      },
      {
        "slug": "claim-2-laplace-poles",
        "title": "The denoiser parameterizes mean evolution in the Laplace domain with stable complex-conjugate poles for direct evaluation at irregular timestamps (Section 4.2).",
        "file": "pages/claim-2-laplace-poles/page.md",
        "children": []
      },
      {
        "slug": "claim-3-gap-aware",
        "title": "A renewal-averaging analysis maps irregular sampling gaps to effective event-domain poles and motivates gap-aware history conditioning (Section 4.3).",
        "file": "pages/claim-3-gap-aware/page.md",
        "children": []
      },
      {
        "slug": "claim-4-forecasting",
        "title": "Across seven real-world datasets, LLapDiff reports the strongest long-horizon probabilistic forecasting results on the main CRPS/MSE comparison (Table 1).",
        "file": "pages/claim-4-forecasting/page.md",
        "children": []
      },
      {
        "slug": "claim-5-imputation",
        "title": "The same queried latent-trajectory model performs missing-value imputation by querying historical timestamps without retraining (Figure 3).",
        "file": "pages/claim-5-imputation/page.md",
        "children": []
      },
      {
        "slug": "conclusion",
        "title": "Conclusion",
        "file": "pages/conclusion/page.md",
        "children": []
      }
    ]
  }
}