tuckergol
commented
6 months ago - Computing bias: bias in algorithms leading to unfair outcomes
- Types of Computing Bias...
- Data Bias: unreflective of real-world values
- Human Bias: personal biases of developers influencing algorithms
- Algorithmic Bias: reflects data and human bias, amplifying it
- Explicit vs. implicit data...
- Explicit Data: information given to an app or site
- Implicit Data: information collected based on user activity
- Mitigation Strategies: obtain more diverse data, counteract biased data
- Pre-processing: clean and organize raw data to check for bias.
- In-processing: manipulate and transform data during active use.
- Post-processing: three-step check and adjustment process after training
- Input Correction: adjust testing data
- Classifier correction: fine-tune the model
- Output correction: modify predictions
Computing Bias | Tucker’s Blog
Computing Bias Team Teach
https://tuckergol.github.io/student2/2023/12/11/Computing-Bias-Team-Teach_IPYNB2.html