2020

Exploring Alternatives to Softmax Function

Banerjee, Kunal, C, Vishak Prasad, Gupta, Rishi Raj et al.

Understand

Softmax function is widely used in artificial neural networks for multiclass classification, multilabel classification, attention mechanisms, etc.

  • However, its efficacy is often questioned in literature.
  • The log-softmax loss has been shown to belong to a more generic class of loss functions, called spherical family, and its member log-Taylor softmax loss is arguably the best alternative in this class.
  • In another approach which tries to enhance the discriminative nature of the softmax function, soft-margin softmax (SM-softmax) has been proposed to be the most suitable alternative.

Built on

  • “ImageNet Large Scale Visual Recognition Challenge”

    Olga Russakovsky et al · 2015

    Earlier work this paper cites.

  • “Efficient Exact Gradient Update for training Deep Networks with Very Large Sparse Targets”

    Pascal Vincent, Alexandre de Brébisson and Xavier Bouthillier · 2015

    Earlier work this paper cites.

  • “An Exploration of Softmax Alternatives Belonging to the Spherical Loss Family”

    Alexandre de Brébisson and Pascal Vincent · 2016

    Earlier work this paper cites.

  • “Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)”

    Original

    Djork-Arné Clevert, Thomas Unterthiner and Sepp Hochreiter · 2016

    Earlier work this paper cites.

Similar

  • “Large-Margin Softmax Loss for Convolutional Neural Networks”

    Weiyang Liu, Yandong Wen, Zhiding Yu and Meng Yang · 2016

    Cited alongside, same era.

  • “Soft-Margin Softmax for Deep Classification”

    Xuezhi Liang et al · 2017

    Cited alongside, same era.

  • “Attention is All you Need”

    Ashish Vaswani et al · 2017

    Cited alongside, same era.

Then

  • “DropMax: Adaptive Variationial Softmax”

    Hae Lee et al · 2018

    Later among the works it cites.

  • “How to Develop a CNN From Scratch for CIFAR-10 Photo Classification” Accessed: 2020-06-21, https://machinelearningmastery.com/how-to-develop-a-cnn-from-scratch-for-cifar-10-photo-classification/

    Jason Brownlee · 2020

    Closest in time.

  • “Show, Attend and Tell: Neural Image Caption Generation with Visual Attention”

    Original

    Kelvin Xu et al · 2057

    Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…