@researai/deepscientist 1.5.16 → 1.6.0

package/AISB/catalog/aisb.t3.078_conformalanomaly.yaml

@@ -0,0 +1,146 @@
+id: aisb.t3.078_conformalanomaly
+name: CADES - Conformal Anomaly Detection in Event Sequences
+version: 0.1.0
+one_line: Statistical anomaly detection framework for continuous-time event sequences
+  with finite-sample FPR control via conformal inference.
+task_description: 'This benchmark entry covers CADES (Conformal Anomaly Detection
+  in Event Sequences), an ICML 2025 method that applies conformal inference to detect
+  anomalous event sequences generated by temporal point processes. The method uses
+  the time-rescaling theorem to design two complementary non-conformity scores that
+  address non-identifiability issues of prior test statistics (e.g., KS statistic,
+  sum-of-squared-spacings). CADES combines these scores with Bonferroni correction
+  and provides both marginal and calibration-conditional FPR guarantees. The current
+  local snapshot preserves paper metadata and metric definitions only; executable
+  source code has not been packaged and requires upstream restoration before evaluation.
+
+  '
+task_mode: analysis_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 2-6h
+cost_band: medium
+difficulty: medium
+data_access: public
+primary_outputs:
+- aid_accuracy
+- auroc_anchorage_ak
+- tpr_at_alpha_005_anchorage_ak
+- auroc_aleutian_islands_ak
+- auroc_helmet_ca
+launch_profiles:
+- id: analysis_only
+  label: Analysis Only
+  description: Review preserved paper metadata, metric contracts, and theoretical
+    framework without executing code.
+- id: restore_first
+  label: Restore Source First
+  description: Upstream restoration required before treating this as a runnable benchmark;
+    no executable code anchors detected in current snapshot.
+dataset_download:
+  primary_method: upstream_restoration
+  sources: []
+  notes:
+  - No bundled dataset route in current snapshot; treat data setup as part of upstream
+    restoration.
+  - Evaluation requires event sequence datasets covering multiple anomaly types per
+    the paper's experimental protocol.
+credential_requirements:
+  mode: none
+  items: []
+  notes:
+  - No credential requirements currently needed for metadata-only snapshot.
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 16
+    disk_gb: 30
+    gpu_count: 1
+    gpu_vram_gb: 8
+  recommended:
+    cpu_cores: 16
+    ram_gb: 32
+    disk_gb: 80
+    gpu_count: 1
+    gpu_vram_gb: 16
+environment:
+  python: '3.8'
+  cuda: '11.3'
+  pytorch: null
+  flash_attn: null
+  key_packages: []
+  notes:
+  - Environment specifications are provisional; no executable setup files or runnable
+    source code preserved in current snapshot.
+  - Full dependency set should be sourced from upstream restoration of the paper's
+    official codebase.
+  - Refer to json/metric_contract.json for metric definitions pending code availability.
+risk_flags:
+- source_snapshot_incomplete
+- route_caveat
+- blocked_no_code
+risk_notes:
+- No executable code files detected in current snapshot; all staged metrics (auroc_anchorage_ak,
+  tpr_at_alpha_005_anchorage_ak, auroc_aleutian_islands_ak, auroc_helmet_ca) are blocked
+  pending source restoration.
+- Do not treat this entry as a ready-to-run benchmark; top-level README absent from
+  snapshot.
+- Metric evaluation requires recovering the paper's official implementation before
+  benchmark execution.
+recommended_when: 'Use this entry for paper-backed planning, literature review of
+  conformal anomaly detection methods, or when contributing upstream restoration of
+  the source code and evaluation pipeline.
+
+  '
+not_recommended_when: 'Do not use this entry as a runnable benchmark; the snapshot
+  preserves metadata only and lacks executable code anchors for the staged metrics.
+  Execution will fail without source restoration.
+
+  '
+paper:
+  title: Conformal Anomaly Detection in Event Sequences
+  venue: ICML 2025
+  year: 2025
+  url: https://proceedings.mlr.press/v267/zhang25dn.html
+  authors:
+  - Shuai Zhang
+  - Chuan Zhou
+  - Yang Liu
+  - Peng Zhang
+  - Xixun Lin
+  - Shirui Pan
+  affiliations:
+  - Academy of Mathematics and Systems Science, Chinese Academy of Sciences
+  - School of Cyber Security, University of Chinese Academy of Sciences
+  - Cyberspace Institute of Advanced Technology, Guangzhou University
+  - Institute of Information Engineering, Chinese Academy of Sciences
+  - Griffith University
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.078_conformalanomaly.zip
+  archive_type: zip
+  local_dir_name: paper-78-ConformalAnomaly
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.078_conformalanomaly.zip
+  sha256: 637c58e105fb132ee3a79aa51c67f7f5fecec8128f04e9245f02a286231dad91
+  size_bytes: 30708
+display:
+  palette_seed: marine-sand-event
+  art_style: temporal-statistics
+  accent_priority: medium
+image_path: ../image/078_aisb.t3.078_conformalanomaly.jpg
+capability_tags:
+- research_code_optimization
+- anomaly_detection
+- event_sequences
+- conformal_prediction
+- temporal_point_processes
+aisb_direction: T3
+track_fit:
+- paper_track
+- benchmark_track
+commercial:
+  annual_fee: null

package/AISB/catalog/aisb.t3.078_conformalanomaly.zh.yaml

@@ -0,0 +1,117 @@
+id: aisb.t3.078_conformalanomaly
+name: CADES - Conformal Anomaly Detection in Event Sequences
+version: 0.1.0
+one_line: 面向连续时间事件序列的统计异常检测框架，基于 conformal inference 提供有限样本 FPR 控制。
+task_description: 该条目覆盖 CADES（Conformal Anomaly Detection in Event Sequences），该 ICML 2025 方法将 conformal inference 用于检测由时序点过程生成的异常事件序列。方法利用 time-rescaling theorem 设计两种互补的非一致性分数，以缓解既有统计量（如 KS、间隔平方和）中的不可辨识问题，并结合 Bonferroni 校正，提供边际与校准条件下的 FPR 保证。当前本地快照仅保留论文元数据与指标定义，不含可执行源码；需先进行上游恢复。
+task_mode: analysis_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 2-6h
+cost_band: medium
+difficulty: medium
+data_access: public
+primary_outputs:
+- aid_accuracy
+- auroc_anchorage_ak
+- tpr_at_alpha_005_anchorage_ak
+- auroc_aleutian_islands_ak
+- auroc_helmet_ca
+launch_profiles:
+- id: analysis_only
+  label: Analysis Only
+  description: Review preserved paper metadata, metric contracts, and theoretical framework without executing code.
+- id: restore_first
+  label: Restore Source First
+  description: Upstream restoration required before treating this as a runnable benchmark; no executable code anchors detected in current snapshot.
+dataset_download:
+  primary_method: upstream_restoration
+  sources: []
+  notes:
+  - No bundled dataset route in current snapshot; treat data setup as part of upstream restoration.
+  - Evaluation requires event sequence datasets covering multiple anomaly types per the paper's experimental protocol.
+credential_requirements:
+  mode: none
+  items: []
+  notes:
+  - No credential requirements currently needed for metadata-only snapshot.
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 16
+    disk_gb: 30
+    gpu_count: 1
+    gpu_vram_gb: 8
+  recommended:
+    cpu_cores: 16
+    ram_gb: 32
+    disk_gb: 80
+    gpu_count: 1
+    gpu_vram_gb: 16
+environment:
+  python: '3.8'
+  cuda: '11.3'
+  pytorch: null
+  flash_attn: null
+  key_packages: []
+  notes:
+  - Environment specifications are provisional; no executable setup files or runnable source code preserved in current snapshot.
+  - Full dependency set should be sourced from upstream restoration of the paper's official codebase.
+  - Refer to json/metric_contract.json for metric definitions pending code availability.
+risk_flags:
+- source_snapshot_incomplete
+- route_caveat
+- blocked_no_code
+risk_notes:
+- No executable code files detected in current snapshot; all staged metrics (auroc_anchorage_ak, tpr_at_alpha_005_anchorage_ak, auroc_aleutian_islands_ak, auroc_helmet_ca) are blocked pending source restoration.
+- Do not treat this entry as a ready-to-run benchmark; top-level README absent from snapshot.
+- Metric evaluation requires recovering the paper's official implementation before benchmark execution.
+recommended_when: 当你进行基于论文的方法规划、共形异常检测文献调研，或准备恢复源码与评估流水线时可使用该条目。
+not_recommended_when: 不要将该条目当作可直接运行的基准；当前仅有元数据，缺少可执行代码锚点，未恢复源码将无法执行。
+paper:
+  title: Conformal Anomaly Detection in Event Sequences
+  venue: ICML 2025
+  year: 2025
+  url: https://proceedings.mlr.press/v267/zhang25dn.html
+  authors:
+  - Shuai Zhang
+  - Chuan Zhou
+  - Yang Liu
+  - Peng Zhang
+  - Xixun Lin
+  - Shirui Pan
+  affiliations:
+  - Academy of Mathematics and Systems Science, Chinese Academy of Sciences
+  - School of Cyber Security, University of Chinese Academy of Sciences
+  - Cyberspace Institute of Advanced Technology, Guangzhou University
+  - Institute of Information Engineering, Chinese Academy of Sciences
+  - Griffith University
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.078_conformalanomaly.zip
+  archive_type: zip
+  local_dir_name: paper-78-ConformalAnomaly
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.078_conformalanomaly.zip
+  sha256: 637c58e105fb132ee3a79aa51c67f7f5fecec8128f04e9245f02a286231dad91
+  size_bytes: 30708
+display:
+  palette_seed: marine-sand-event
+  art_style: temporal-statistics
+  accent_priority: medium
+image_path: ../image/078_aisb.t3.078_conformalanomaly.jpg
+capability_tags:
+- research_code_optimization
+- anomaly_detection
+- event_sequences
+- conformal_prediction
+- temporal_point_processes
+aisb_direction: T3
+track_fit:
+- paper_track
+- benchmark_track
+commercial:
+  annual_fee: null

package/AISB/catalog/aisb.t3.079_dpfkmeans.yaml

@@ -0,0 +1,131 @@
+id: aisb.t3.079_dpfkmeans
+name: Differentially Private Federated k-Means Clustering with Server-Side Data
+version: 0.1.0
+one_line: Federated k-means clustering with differential privacy guarantees using
+  server-side auxiliary data for initialization.
+task_description: 'This benchmark covers FedDP-KMeans, a fully federated and differentially
+  private k-means clustering algorithm. The method combines FedDP-Init, a novel initialization
+  approach that leverages potentially small and out-of-distribution server-side data,
+  with FedDP-Lloyd''s, a federated variant of Lloyd''s algorithm. The approach addresses
+  the primary challenge in differentially private clustering: obtaining good initialization
+  under privacy constraints. The local snapshot currently requires upstream source
+  restoration before it becomes runnable; no executable code anchors were detected
+  in the current staging pass.
+
+  '
+task_mode: analysis_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 2-6h
+cost_band: low
+difficulty: medium
+data_access: public
+primary_outputs:
+- objective_score
+launch_profiles:
+- id: analysis_only
+  label: Analysis Only
+  description: Inspect preserved benchmark metadata for the differentially private
+    federated k-means task without attempting execution.
+- id: restore_first
+  label: Restore First
+  description: Restore missing clustering implementation (FedDP-Init and FedDP-Lloyd's)
+    before attempting benchmark execution.
+dataset_download:
+  primary_method: restoration_required
+  sources:
+  - type: upstream_restore
+    location: github.com/jonnyascott/fed-dp-kmeans
+    description: Original implementation repository requires restoration to local
+      snapshot
+  notes:
+  - Current snapshot does not preserve an executable benchmark route; treat dataset
+    and code setup as part of upstream restoration
+  - Public datasets referenced in paper include synthetic and real-world benchmarks
+    for cross-silo and cross-device FL settings
+credential_requirements:
+  mode: none
+  items: []
+  notes:
+  - Benchmark targets public data scenarios with differential privacy guarantees
+  - No external API keys or private data access required
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 16
+    disk_gb: 50
+    gpu_count: 0
+    gpu_vram_gb: 0
+  recommended:
+    cpu_cores: 16
+    ram_gb: 32
+    disk_gb: 100
+    gpu_count: 0
+    gpu_vram_gb: 0
+environment:
+  python: null
+  cuda: null
+  pytorch: 2.4.0
+  flash_attn: null
+  key_packages:
+  - numpy
+  - scipy
+  - scikit-learn
+  notes:
+  - CPU-only execution is plausible for the minimum configuration route
+  - Current package does not preserve executable setup files; environment hints are
+    provisional pending upstream restoration
+  - Original implementation uses standard scientific Python stack; see bundled README/requirements
+    when restored
+risk_flags:
+- source_snapshot_incomplete
+- blocked_no_code
+- route_caveat
+risk_notes:
+- Current snapshot has no executable code anchors; primary_outputs metric (objective_score)
+  is blocked-no-code status
+- No benchmark execution was performed in this packaging pass
+- Top-level README is absent in the current snapshot
+- Upstream restoration from github.com/jonnyascott/fed-dp-kmeans is required before
+  evaluation
+recommended_when: 'Use this entry when conducting paper-backed research planning,
+  auditing the ICML 2025 FedDP-KMeans publication for methodology details, or performing
+  upstream restoration of the differential-privacy clustering implementation.
+
+  '
+not_recommended_when: 'Do not treat this local package as a ready-to-run clustering
+  benchmark. The executable code is absent from the staged snapshot; execution attempts
+  will fail without prior restoration from the upstream GitHub repository.
+
+  '
+paper:
+  title: Differentially Private Federated k-Means Clustering with Server-Side Data
+  authors:
+  - Jonathan Scott
+  - Christoph H. Lampert
+  - David Saulpic
+  affiliations:
+  - Institute of Science and Technology Austria (ISTA)
+  - CNRS & Université Paris Cité
+  venue: ICML 2025
+  year: 2025
+  url: https://arxiv.org/abs/2506.05408
+  code_url: https://github.com/jonnyascott/fed-dp-kmeans
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.079_dpfkmeans.zip
+  archive_type: zip
+  local_dir_name: paper-79-DPFKMeans
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.079_dpfkmeans.zip
+  sha256: 53ac3827f51dfe09b4dec7226640cfd0eea7f0ee2829abff4aeca43f11963ab3
+  size_bytes: 49426
+display:
+  palette_seed: plum-ivory-privacy
+  art_style: privacy-clustering
+  accent_priority: medium
+image_path: ../image/079_aisb.t3.079_dpfkmeans.jpg

package/AISB/catalog/aisb.t3.079_dpfkmeans.zh.yaml

@@ -0,0 +1,104 @@
+id: aisb.t3.079_dpfkmeans
+name: Differentially Private Federated k-Means Clustering with Server-Side Data
+version: 0.1.0
+one_line: 带差分隐私保证的联邦 k-means 聚类方法，利用服务器侧辅助数据进行初始化。
+task_description: 该基准覆盖 FedDP-KMeans，一种完全联邦且满足差分隐私的 k-means 算法。方法将 FedDP-Init（利用可能较小且分布外的服务器侧数据进行初始化）与 FedDP-Lloyd’s（Lloyd 算法的联邦变体）结合，解决差分隐私聚类中的核心难点：在隐私约束下获得高质量初始化。当前本地快照需先恢复上游源码后方可运行；本次分阶段打包未检测到可执行代码锚点。
+task_mode: analysis_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 2-6h
+cost_band: low
+difficulty: medium
+data_access: public
+primary_outputs:
+- objective_score
+launch_profiles:
+- id: analysis_only
+  label: Analysis Only
+  description: Inspect preserved benchmark metadata for the differentially private federated k-means task without attempting execution.
+- id: restore_first
+  label: Restore First
+  description: Restore missing clustering implementation (FedDP-Init and FedDP-Lloyd's) before attempting benchmark execution.
+dataset_download:
+  primary_method: restoration_required
+  sources:
+  - type: upstream_restore
+    location: github.com/jonnyascott/fed-dp-kmeans
+    description: Original implementation repository requires restoration to local snapshot
+  notes:
+  - Current snapshot does not preserve an executable benchmark route; treat dataset and code setup as part of upstream restoration
+  - Public datasets referenced in paper include synthetic and real-world benchmarks for cross-silo and cross-device FL settings
+credential_requirements:
+  mode: none
+  items: []
+  notes:
+  - Benchmark targets public data scenarios with differential privacy guarantees
+  - No external API keys or private data access required
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 16
+    disk_gb: 50
+    gpu_count: 0
+    gpu_vram_gb: 0
+  recommended:
+    cpu_cores: 16
+    ram_gb: 32
+    disk_gb: 100
+    gpu_count: 0
+    gpu_vram_gb: 0
+environment:
+  python: null
+  cuda: null
+  pytorch: 2.4.0
+  flash_attn: null
+  key_packages:
+  - numpy
+  - scipy
+  - scikit-learn
+  notes:
+  - CPU-only execution is plausible for the minimum configuration route
+  - Current package does not preserve executable setup files; environment hints are provisional pending upstream restoration
+  - Original implementation uses standard scientific Python stack; see bundled README/requirements when restored
+risk_flags:
+- source_snapshot_incomplete
+- blocked_no_code
+- route_caveat
+risk_notes:
+- Current snapshot has no executable code anchors; primary_outputs metric (objective_score) is blocked-no-code status
+- No benchmark execution was performed in this packaging pass
+- Top-level README is absent in the current snapshot
+- Upstream restoration from github.com/jonnyascott/fed-dp-kmeans is required before evaluation
+recommended_when: 当你进行基于论文的方法规划、审阅 ICML 2025 FedDP-KMeans 技术细节，或执行差分隐私聚类实现的上游恢复时可使用本条目。
+not_recommended_when: 不要将当前本地包视为可直接运行的聚类基准。分阶段快照缺少可执行代码，恢复上游仓库前执行会失败。
+paper:
+  title: Differentially Private Federated k-Means Clustering with Server-Side Data
+  authors:
+  - Jonathan Scott
+  - Christoph H. Lampert
+  - David Saulpic
+  affiliations:
+  - Institute of Science and Technology Austria (ISTA)
+  - CNRS & Université Paris Cité
+  venue: ICML 2025
+  year: 2025
+  url: https://arxiv.org/abs/2506.05408
+  code_url: https://github.com/jonnyascott/fed-dp-kmeans
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.079_dpfkmeans.zip
+  archive_type: zip
+  local_dir_name: paper-79-DPFKMeans
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.079_dpfkmeans.zip
+  sha256: 53ac3827f51dfe09b4dec7226640cfd0eea7f0ee2829abff4aeca43f11963ab3
+  size_bytes: 49426
+display:
+  palette_seed: plum-ivory-privacy
+  art_style: privacy-clustering
+  accent_priority: medium
+image_path: ../image/079_aisb.t3.079_dpfkmeans.jpg

package/AISB/catalog/aisb.t3.080_latentscorereweight.yaml

@@ -0,0 +1,169 @@
+schema_version: 1
+id: aisb.t3.080_latentscorereweight
+name: Latent Score-Based Reweighting for Robust Classification on Imbalanced Tabular
+  Data
+version: 0.1.0
+one_line: 'Score-based density reweighting for robust tabular classification under
+  distribution shift, without requiring group labels.
+
+  '
+task_description: 'This packaged benchmark targets robust tabular classification by
+  identifying and upweighting underrepresented regions of the joint data distribution
+  P(X, Y) using latent score signals from a score-based (diffusion) model. Unlike
+  boundary-based methods such as JTT that upweight misclassified samples near the
+  decision boundary P(Y|X), this framework estimates local sample density via score-vector
+  similarity with neighbouring data points, then reweights low-density regions to
+  produce a more globally balanced training distribution. The method requires no group-label
+  prior information. Evaluation targets worst-group accuracy and mean accuracy on
+  tabular datasets under distribution shift. The current local snapshot does not preserve
+  executable training or evaluation code; both primary metrics are blocked-no-code
+  and the entry must be treated as restore-needed before any run is attempted.
+
+  '
+task_mode: analysis_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 6-24h
+cost_band: medium
+difficulty: medium
+data_access: public
+capability_tags:
+- robust_classification
+- tabular_ml
+- score_based_models
+- distribution_shift
+- data_reweighting
+- imbalanced_data
+aisb_direction: T3
+track_fit:
+- paper_track
+- benchmark_track
+primary_outputs:
+- worst_group_accuracy
+- mean_accuracy
+launch_profiles:
+- id: analysis_only
+  label: Analysis Only
+  description: 'Review preserved benchmark metadata, metric contracts, and paper artefacts.
+    No executable code is present in the current snapshot; this profile supports planning
+    and paper-backed analysis only.
+
+    '
+- id: restore_first
+  label: Restore First
+  description: 'Restore the missing latent-score reweighting implementation from the
+    upstream repository (https://github.com/YunzeTong/latent-score-based-reweighting)
+    before treating this entry as runnable. Both metrics remain blocked until restoration.
+
+    '
+dataset_download:
+  primary_method: mixed
+  sources: []
+  notes:
+  - 'The current local snapshot does not preserve an executable benchmark route. Dataset
+    setup should be treated as part of upstream restoration rather than a bundled
+    run path. Refer to the paper and upstream GitHub repository for dataset identities
+    and loading instructions.
+
+    '
+credential_requirements:
+  mode: none
+  items: []
+  notes: []
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 32
+    disk_gb: 80
+    gpu_count: 1
+    gpu_vram_gb: 16
+  recommended:
+    cpu_cores: 16
+    ram_gb: 64
+    disk_gb: 150
+    gpu_count: 1
+    gpu_vram_gb: 24
+environment:
+  python: null
+  cuda: null
+  pytorch: null
+  flash_attn: null
+  key_packages: []
+  notes:
+  - 'The current local snapshot does not preserve executable setup files or runnable
+    source code. Environment requirements should be treated as provisional until upstream
+    restoration from https://github.com/YunzeTong/latent-score-based-reweighting is
+    complete.
+
+    '
+  - 'Expected dependencies include a score-based or diffusion model library, standard
+    tabular ML packages (e.g. scikit-learn, XGBoost), and a PyTorch backend for the
+    generative density estimation stage. Confirm exact versions from the upstream
+    repository''s requirements file after restoration.
+
+    '
+risk_flags:
+- source_snapshot_incomplete
+- route_caveat
+- metrics_blocked_no_code
+risk_notes:
+- 'Both primary metrics (worst_group_accuracy, mean_accuracy) are marked blocked-no-code
+  in the metric contract. No executable code anchor was detected in the staged snapshot.
+  Do not report metric values until trusted execution outputs exist from a restored
+  environment.
+
+  '
+- 'The top-level README is absent from the current snapshot. Do not rely on README-only
+  interpretation; prefer the paper (latex.md) and metric contract (json/metric_contract.json)
+  as authoritative references.
+
+  '
+recommended_when: 'Use this entry when you want paper-backed planning, method analysis,
+  or upstream restoration scoping for the latent score-based reweighting robustness
+  approach. Also appropriate when comparing density-based reweighting strategies against
+  boundary-based baselines (e.g. JTT) at the design or literature-review stage.
+
+  '
+not_recommended_when: 'Do not treat this local package as a ready-to-run robustness
+  benchmark. The executable generative reweighting route is absent from the staged
+  snapshot and both metrics are blocked. Any reported numbers must come from a fully
+  restored and verified execution environment, not from this package in its current
+  state.
+
+  '
+paper:
+  title: Latent Score-Based Reweighting for Robust Classification on Imbalanced Tabular
+    Data
+  authors:
+  - Yunze Tong
+  - Fengda Zhang
+  - Zihao Tang
+  - Kaifeng Gao
+  - Kai Huang
+  - Pengfei Lyu
+  - Jun Xiao
+  - Kun Kuang
+  venue: ICML 2025
+  year: 2025
+  url: https://proceedings.mlr.press/v267/tong25c.html
+  code_url: https://github.com/YunzeTong/latent-score-based-reweighting
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.080_latentscorereweight.zip
+  archive_type: zip
+  local_dir_name: paper-80-LatentScoreReweight
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.080_latentscorereweight.zip
+  sha256: 9c70bcc4a57a17ffc712a81763731ba5c5f46c987ca83e9b5cd0d730a18e0d03
+  size_bytes: 36629
+commercial:
+  annual_fee: null
+display:
+  palette_seed: copper-teal-reweight
+  art_style: robust-tabular
+  accent_priority: high
+image_path: ../image/080_aisb.t3.080_latentscorereweight.jpg