File size: 3,122 Bytes
90d9fcf b295e5b | 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 60 61 62 63 64 65 66 67 68 69 | {
"schema_version": 1,
"title": "Reproduction: Adaptive Estimation and Inference in Semi-parametric Heterogeneous Clustered Multitask Learning via Neyman Orthogonality",
"emoji": "🧮",
"space_id": "snaykey/repro-profiling-irrational-agent",
"paper": {
"openreview_id": "5hDvooOKUP"
},
"tags": [
"icml2026-repro",
"paper-5hDvooOKUP"
],
"updated_at": "2026-07-31T00:00:00Z",
"root": {
"slug": "index",
"title": "Reproduction: Adaptive Estimation and Inference in Semi-parametric Heterogeneous Clustered Multitask Learning via Neyman Orthogonality",
"file": "pages/index.md",
"children": [
{
"slug": "executive-summary",
"title": "Executive summary",
"file": "pages/executive-summary/page.md",
"children": []
},
{
"slug": "claim-1-thm35-cluster-recovery",
"title": "The adaptive orthogonal multitask estimator achieves exact recovery of the latent task clustering with high probability, established via Theorem 3.5 (Section 3, Cluster Recovery).",
"file": "pages/claim-1-thm35-cluster-recovery/page.md",
"children": []
},
{
"slug": "claim-2-thm35-pooled-rate",
"title": "For tasks in cluster k with pooled sample size N_k, the estimator attains the rate ||θ̂_j − θ*_j||_2 = O_P(N_k^{-1/2}), matching pooled parametric convergence (Theorem 3.5, Section 3).",
"file": "pages/claim-2-thm35-pooled-rate/page.md",
"children": []
},
{
"slug": "claim-3-thm36-asymptotic-normality",
"title": "√N_k(θ̂_j − θ*_j) is asymptotically normal with covariance matching the oracle estimator that knows the true clustering in advance (Theorem 3.6, Section 3).",
"file": "pages/claim-3-thm36-asymptotic-normality/page.md",
"children": []
},
{
"slug": "claim-4-thm37-38-within-cluster-heterogeneity",
"title": "Extensions permit within-cluster heterogeneity bounded by ξ_k = O(N_k^{-1/2}) while preserving the estimation guarantees (Theorems 3.7–3.8, Section 3).",
"file": "pages/claim-4-thm37-38-within-cluster-heterogeneity/page.md",
"children": []
},
{
"slug": "claim-5-sec44-simulations-ari",
"title": "In simulations across three models (PLM, ATE, DID) and separation levels δ ∈ {1/3, 2/3, 1}, the method achieves Adjusted Rand Index (ARI) near 1, outperforming competing clustering approaches (Section 4.4).",
"file": "pages/claim-5-sec44-simulations-ari/page.md",
"children": []
},
{
"slug": "claim-6-sec5-recs-real-data",
"title": "On real electricity price elasticity data from 51 US states, the method recovers three clusters, e.g. Virginia at −1.138 ± 0.189 versus a 46-state cluster at −0.221 ± 0.009 (Table 1, Section 5).",
"file": "pages/claim-6-sec5-recs-real-data/page.md",
"children": []
},
{
"slug": "conclusion",
"title": "Conclusion",
"file": "pages/conclusion/page.md",
"children": []
}
]
}
} |