admin管理员组

文章数量:1206772

I wish to use a vector-valued activation function in my neural network which is injective that is, maps inputs to unique outputs. I require this property to maintain injectivity of my neural network. This has several use cases as it would make the network invertible and the resulting function that the network represents would also have properties that may be required for domain specific downstream tasks.

Are there any injective vector-valued activation functions?

Can any of the normalizations be made injective? In particular, I am using my network to evaluate a function at different x values so a normalization like layer normalization makes more sense.

本文标签： deep learningInjective vectorvalued activation functions Injective normalizationStack Overflow

版权声明：本文标题：deep learning - Injective vector-valued activation functions. Injective normalization? - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://www.betaflare.com/web/1738726950a2109095.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。