all AI news
If Influence Functions are the Answer, Then What is the Question?. (arXiv:2209.05364v1 [cs.LG])
Sept. 13, 2022, 1:13 a.m. | Juhan Bae, Nathan Ng, Alston Lo, Marzyeh Ghassemi, Roger Grosse
stat.ML updates on arXiv.org arxiv.org
Influence functions efficiently estimate the effect of removing a single
training data point on a model's learned parameters. While influence estimates
align well with leave-one-out retraining for linear models, recent works have
shown this alignment is often poor in neural networks. In this work, we
investigate the specific factors that cause this discrepancy by decomposing it
into five separate terms. We study the contributions of each term on a variety
of architectures and datasets and how they vary with factors …
More from arxiv.org / stat.ML updates on arXiv.org
Jobs in AI, ML, Big Data
Senior ML Researcher - 3D Geometry Processing | 3D Shape Generation | 3D Mesh Data
@ Promaton | Europe
Principal Data Engineer
@ RS21 | Remote
SQL/Power BI Developer
@ ICF | Virginia Remote Office (VA99)
Senior Machine Learning Engineer (Canada Remote)
@ Fullscript | Ottawa, ON
Software Engineer - MLOps.
@ Renesas Electronics | Toyosu, Japan
Junior Data Scientist / Artificial Intelligence consultant
@ Deloitte | Luxembourg, LU