x

tony

Message

85

3y

Practical Pitfalls of Causal Scrubbing

TL;DR: We evaluate Causal Scrubbing (CaSc) on synthetic graphs with known ground truth to determine its reliability in confirming correct hypotheses and rejecting incorrect ones. First, we show that CaSc can accurately identify true hypotheses and quantify the degree to which a hypothesis is wrong. Second, we highlight some limitations...

Mar 27, 202387

tony

Subscribe

Message

85

3y

tony

Practical Pitfalls of Causal Scrubbing

TL;DR: We evaluate Causal Scrubbing (CaSc) on synthetic graphs with known ground truth to determine its reliability in confirming correct hypotheses and rejecting incorrect ones. First, we show that CaSc can accurately identify true hypotheses and quantify the degree to which a hypothesis is wrong. Second, we highlight some limitations...

Mar 27, 202387

Practical Pitfalls of Causal Scrubbing

Jérémy Scheurer

Jérémy Scheurer, Phil3, tony, jacquesthibs, David Lindner

3y

TL;DR: We evaluate Causal Scrubbing (CaSc) on synthetic graphs with known ground truth to determine its reliability in confirming correct hypotheses and rejecting incorrect ones. First, we show that CaSc can accurately identify true hypotheses and quantify the degree to which a hypothesis is wrong. Second, we highlight some limitations of CaSc, in particular, that it cannot falsify all incorrect hypotheses. We provide concrete examples of false positive results with causal scrubbing. Our main finding is that false positives can occur when there is “cancellation”, i.e., CaSc causes the model to do better on some inputs and worse on others, such that, on average, the scrubbed model recovers the full loss. A... (read 3713 more words →)

17

87

LESSWRONG
LW

LESSWRONG
LW

tony

tony

tony

Practical Pitfalls of Causal Scrubbing

tony

tony

tony

Practical Pitfalls of Causal Scrubbing