Temporal Predictive Multiplicity: Equally Accurate Time Series Models Yield Different Forecast Trajectories
Emanuele Albini, Francesca Toni, Saumitra Mishra, Francesco Leofante
実装難易度
Hard
推論・学習コスト
High
想定用途
予測
概要
Models with near-identical predictive performance can yield substantially different predictions, a phenomenon known as predictive multiplicity. Prior work has mostly studied this at the level of individual scalar outputs. In time-series forecasting, however, predictions across horizons jointly define a trajectory, and horizon-wise comparisons can hide important differences in predictive behavior.…
何が新しいか
Models with near-identical predictive performance can yield substantially different predictions, a phenomenon known as predictive multiplicity. Prior work has mostly studied this at the level of…
何に使えるか
予測
実装情報
- Paper URL
- あり
実装チェックリスト
実装または配布ページ
要確認Paper onlyの可能性があるため再実装前提で確認してください。
一次情報リンク
OKPaper
検証しやすさ
要確認公式実装が見つからないため、論文から再実装する前提です。
計算資源
要確認学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
ライセンス
未取得配布元のLICENSE、モデルカード、Paperの利用条件を確認してください。
商用利用
未取得研究利用限定、データセット由来制限、API規約の有無を確認してください。
自社データで試すなら
製造業・材料開発のExcel/CSVデータに落とし込むための最初の手順です。
- 1まず自社データを、入力条件、目的変数、評価したい指標に分けて整理します。
- 2正常データだけで動くベースラインを作り、異常スコアのしきい値を現場知見と合わせます。
- 3評価指標はR2/RMSE、AUC、異常検知の再現率、実験回数削減率など、現場の意思決定に近いものを選びます。
- 4SHAPや特徴量重要度で、効いている因子が物理・化学・工程知識と矛盾しないか確認します。
実装難易度
Hard - 公式実装が見つからないため、論文から再実装する前提です。
必要リソース
- GPU目安: High
- データセット: 論文・リポジトリ側の指定を確認してください。
- 学習要否: 再学習や評価環境の準備が必要になる可能性があります。
- 学習や高解像度推論ではGPUメモリと実行時間に注意が必要です。
実務で使う場合の注意点
- ライセンスと商用利用条件は、Paper / GitHub / Hugging Face の配布元で確認してください。
- 精度、再現性、計算コストはデータセットや評価条件に依存します。
- 個人情報や機密データを扱う場合は、入力データの保存先と外部API利用条件を確認してください。
関連記事
Evaluating Sequence Assembly Strategies for Differentially Private Synthetic Time-Series Forecasting
Differentially private time-series generators commonly produce fixed-length synthetic windows, whereas downstr
Prediction-powered inference for time series across space
The following motif is common in spatiotemporal settings: we have a sequence of covariate and label pairs obse
Temporal Visuo-Tactile Learning for Dexterous Grasp Stability
Humans can grasp everyday objects with almost perfect success rates using fingertip tactile feedback, yet much
Identifiability of a dissipative knowledge-dynamics model: exact recovery under designed excitation, degeneration on observational data
Human learning is a dissipative dynamical process: mastery accumulates through practice, decays through forget