Build your first system quickly, then iterate

If you’re working on a brand new machine learning application, you should build your first system quickly and then iterate.

Speech recognition example

There are specific techniques for making speech recognition systems more robust to noisy background.

• Noisy background
  • cafe noise
  • car noise
• Accented Speech
• Far from microphon
• Young shildren’s Speech
• Stuttering

Even experts still find it maybe a little bit difficult to pick without spending some time thinking about the problem.

So, build a system quickly and then iterate.

1. Set up dev/test set and metric.
2. Build initial system quickly
3. Use bias/variance analysis & error analysis to prioritize next steps.