{
  "project": "causal-regression",
  "title": "Causal Regression: Learning Causal Mechanisms for Robust and Interpretable Prediction",
  "public_url": "https://research.wehub.us/ghy/tmp/causal-regression/",
  "progress_url": "https://research.wehub.us/ghy/tmp/causal-regression/progress.json",
  "last_updated": "2026-05-30 17:49 CST",
  "status": "active_validation_extension",
  "current_focus": "Selective archive and public review surface for Causal Regression, the robust-prediction validation line downstream of DiscoSCM.",
  "canonical_seed": "Causal Regression shifts prediction from fitting X->Y associations to abducting latent causes U and predicting through Y=f(U, epsilon).",
  "canonical_source": {
    "host": "gongqian",
    "source_root": "/Users/gongqian/.openclaw/workspace/.me/research/causual_regression/submitted/iclr2026",
    "main_tex": "/Users/gongqian/.openclaw/workspace/.me/research/causual_regression/submitted/iclr2026/main_final_rebuttal.tex",
    "path_note": "upstream directory currently spells causal as causual_regression",
    "main_tex_sha256": "35d7747a77cb257112a0b1ae09c70cfb355a5478f1bee05b15b5bd1e90c98691",
    "main_tex_mtime": "2026-03-19T11:00:06"
  },
  "canonical_source_sha256": "35d7747a77cb257112a0b1ae09c70cfb355a5478f1bee05b15b5bd1e90c98691",
  "source_policy": "selected_metadata_and_core_content_only_no_full_latex_copy",
  "local_extraction_artifact": "/Users/cms/.openclaw/workspace/projects/ghy-temporary-research-archive/outputs/causal-regression-source-extract-v0.1.json",
  "anonymous_code_url": "https://anonymous.4open.science/r/causal-regression-135C",
  "milestones": [
    {
      "id": "cr1",
      "label": "Inspect existing ghy/tmp archive structure",
      "status": "completed"
    },
    {
      "id": "cr2",
      "label": "Extract needed metadata/core content from submitted LaTeX",
      "status": "completed"
    },
    {
      "id": "cr3",
      "label": "Create archive docs and public HTML/progress JSON",
      "status": "completed_pending_deploy"
    },
    {
      "id": "cr4",
      "label": "Deploy and verify public page",
      "status": "completed",
      "completed_at": "2026-05-30 15:43 CST"
    },
    {
      "id": "cr5",
      "title": "Math/diagram page optimization",
      "status": "completed",
      "completed_at": "2026-05-30 15:48 CST",
      "notes": "Clarified mathematical spine and added inline SVG/HTML diagrams for theory shift, CausalEngine pipeline, uncertainty decomposition, and paper-claim result bars."
    },
    {
      "id": "cr6",
      "title": "Diagram arrow visual polish",
      "status": "completed",
      "completed_at": "2026-05-30 16:06 CST",
      "notes": "Replaced blunt/dashed arrows with curved gradient arrow paths, softer glow underlays, rounded arrowheads, and a cleaner compare transition arrow."
    },
    {
      "id": "cr7",
      "title": "MathJax display repair",
      "status": "completed",
      "completed_at": "2026-05-30 16:32 CST",
      "notes": "Split the hero core-shift display formula into aligned lines, constrained display MathJax overflow, and removed a diagram-scoped MathJax SVG override that could interfere with formula rendering."
    },
    {
      "id": "cr8",
      "title": "Diagram arrow scale polish",
      "status": "completed",
      "updated_at": "2026-05-30 17:49 CST",
      "summary": "Used Codex CLI to shrink the SVG arrow visual system: smaller arrowheads, thinner flow strokes, softer glow, and tighter selectors that do not affect MathJax.",
      "verification": [
        "Codex completed the source patch in site/ghy/tmp/causal-regression/index.html",
        "markerWidth/markerHeight reduced from 12 to 10 across custom SVG arrows",
        "flow-arrow and flow-glow stroke widths reduced across diagrams",
        "global .flow-* selectors tightened to diagram/arrow-mid SVG scope"
      ]
    }
  ],
  "validation": {
    "local_extraction_json_exists": true,
    "canonical_main_sha256_recorded": true,
    "full_latex_tree_copied": false,
    "public_html_created": true,
    "public_json_created": true,
    "deployed": true,
    "experiment_reproduction_status": "not_reproduced_from_code_yet",
    "public_http_verified": true,
    "public_html_verified": true,
    "public_json_verified": true,
    "page_redesign": "math-display-repaired-v4-ready-for-deploy"
  },
  "review_rule": "This page is a noindex temporary review projection. Treat the remote submitted TeX as source-of-truth; this archive stores only selected metadata/core content.",
  "core_content": {
    "abstract": "The performance of standard regression models, which primarily learn statistical associations, is vulnerable to label noise. This paper proposes Causal Regression, a paradigm that shifts the focus toward learning invariant causal mechanisms. We introduce CausalEngine, a neural architecture that operationalizes this paradigm based on the Distribution-consistency Structural Causal Model (DiscoSCM). It first performs abduction to infer a distribution over latent cause, and subsequently applies a causal mechanism to make a prediction. The mathematical properties of the Cauchy distribution facilitate an analytical inference process. This design sidesteps the need for sampling-based approximations, thereby eliminating the high-variance gradients and computational overhead they introduce, leading to stable and efficient end-to-end training. This design also provides a structured form of interpretability by decomposing predictive uncertainty into two distinct sources: epistemic uncertainty, arising from incomplete knowledge of an individual, and aleatoric uncertainty, stemming from inherent environmental randomness. Our experiments demonstrate CausalEngine's significant robustness against label noise. Especially in high-noise regimes where strong baselines falter, our approach exhibits a significantly smaller drop in performance. This work suggests that shifting the modeling focus from statistical associations to causal structures is a promising direction for building AI systems that are more reliable and interpretable.Code is available at https://anonymous.4open.science/r/causal-regression-135C.",
    "main_equation": "Y = f(U, epsilon)",
    "framework": [
      "Abduction: infer P(U|X) or P(U|Z) from observed features/representation.",
      "Causal mechanism: apply an invariant f from latent cause U and noise epsilon to prediction Y.",
      "Uncertainty split: U carries epistemic uncertainty; epsilon carries aleatoric uncertainty."
    ],
    "causal_engine_stages": [
      "Perception",
      "Abduction",
      "Action",
      "Decision"
    ],
    "technical_choices": [
      "Cauchy distribution",
      "linear causal mechanism",
      "analytical uncertainty propagation"
    ],
    "experiment_claims_snapshot": [
      "Synthetic and public regression evaluation under label noise.",
      "Noise types: shuffle, outlier, asymmetric, systematic.",
      "Primary metric: Median Absolute Error (MdAE).",
      "High-noise robustness claims are extracted from the paper and are not independently reproduced yet."
    ]
  },
  "source_files": [
    {
      "file": "main_final_rebuttal.tex",
      "size_bytes": 5004,
      "lines": 107,
      "sha256": "35d7747a77cb257112a0b1ae09c70cfb355a5478f1bee05b15b5bd1e90c98691",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "math_commands.tex",
      "size_bytes": 12284,
      "lines": 508,
      "sha256": "90473c4d0542070db244cea73ef962d6cddc5b2a746757e6a40ddf5fdfb90ba9",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_introduction.tex",
      "size_bytes": 18178,
      "lines": 127,
      "sha256": "feb6f1ef0090bd9fa1822a5def91377b342de42b1ea1488f37fb247004fcdbfc",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_causalregression.tex",
      "size_bytes": 4890,
      "lines": 36,
      "sha256": "4f2e849fb0e29db9ee25c9fb58da054a4a9c3545c63595fc41f75ae540348164",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_causalengine.tex",
      "size_bytes": 12920,
      "lines": 99,
      "sha256": "3a4cbe696abe4ba9032b3abfd6076f837fb6a606c8254b6d8c51f939c5d9d569",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_experiments.tex",
      "size_bytes": 11762,
      "lines": 154,
      "sha256": "aadabd2257824b8f40472b38aca68322f787cb4432e6c630eac0661a75945f7c",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_conclusion.tex",
      "size_bytes": 3684,
      "lines": 28,
      "sha256": "f73224a0bfe86791da1607956a7c31e185bf13bd6f3484c12983c88de6259a6d",
      "mtime": "2026-03-19T11:00:06"
    },
    {
      "file": "final_appendix.tex",
      "size_bytes": 35060,
      "lines": 427,
      "sha256": "d45412189de33142fec74e61dfb1339859460468a30a836c76631409aff09a77",
      "mtime": "2026-03-19T11:00:06"
    }
  ],
  "pdf_variants": [
    {
      "file": "7356_Causal_Regression_Learnin.pdf",
      "size_bytes": 8379869,
      "mtime": "2026-03-19T11:01:24"
    },
    {
      "file": "iclr2026_conference.pdf",
      "size_bytes": 8337063,
      "mtime": "2026-03-19T11:01:24"
    },
    {
      "file": "main_final.pdf",
      "size_bytes": 8289301,
      "mtime": "2026-03-19T11:01:24"
    },
    {
      "file": "main_final_rebuttal.pdf",
      "size_bytes": 8289301,
      "mtime": "2026-03-19T11:01:25"
    },
    {
      "file": "main_full_arxiv.pdf",
      "size_bytes": 8339760,
      "mtime": "2026-03-19T11:01:24"
    }
  ],
  "urls": [
    "https://anonymous.4open.science/r/causal-regression-135C"
  ]
}