Interpretable machine learning applications: Part 5

提供:
Coursera Project Network
このガイド付きプロジェクトでは、次のことを行います。

 Be acquainted with the basics of the Aequitas Tool as a tool to measure and detect bias in the outcome of a machine learning prediction model.

Learn more about a real world case study, i.e., predictions of recidivism (COMPAS dataset), and how the prediction model may have been biased.

Learn a technique, which is largely based on statistical descriptors, for measuring bias and fairness for Machine Learning (ML) prediction models.

Clock1.5 hours
Beginner初級
Cloudダウンロード不要
Video分割画面ビデオ
Comment Dots英語
Laptopデスクトップのみ

You will be able to use the Aequitas Tool as a tool to measure and detect bias in the outcome of a machine learning prediction model. As a use case, we will be working with the dataset about recidivism, i.e., the likelihood for a former imprisoned person to commit another offence within the first two years, since release from prison. The guided project will be making use of the COMPAS dataset, which already includes predicted as well as actual outcomes. Given also that this technique is largely based on statistical descriptors for measuring bias and fairness, it is very independent from specific Machine Learning (ML) prediction models. In this sense, the project will boost your career not only as a Data Scientists or ML developer, but also as a policy and decision maker.

あなたが開発するスキル

  • Software Engineering
  • Artificial Intelligence (AI)
  • Data Science

ステップバイステップで学習します

ワークエリアを使用した分割画面で再生するビデオでは、講師がこれらの手順を説明します。

  1. Setting up the stage

  2. First attempt and stage for detecting bias

  3. Second attempt and stage for detecting bias

  4. Third attempt and stage in detecting bias

  5. Visualisation: Final stage for detecting bias

ガイド付きプロジェクトの仕組み

ワークスペースは、ブラウザに完全にロードされたクラウドデスクトップですので、ダウンロードは不要です

分割画面のビデオで、講師が手順ごとにガイドします

よくある質問

よくある質問

さらに質問がある場合は、受講者ヘルプセンターにアクセスしてください。