“13 NIPS Papers that caught our eye”

Zygmunt Zając at fastml.com has a great post titled “13 NIPS Papers that caught our eye.” Zajac provides a short readable summary of his favorite NIPS 2013 papers.  (NIPS 2013 which just ended last week.) The papers are:

  1. Understanding Dropout  by Baldi and Sadowski
  2. Training and Analysing Deep Recurrent Neural Networks by Hermans and Schrauwen
  3. RNADE: The real-valued neural autoregressive density-estimator by Uria, Murray, and Larochelle
  4. Predicting Parameters in Deep Learning by Denil, Shakibi, Dinh,  Ranzato, and Freitas
  5. Pass-efficient unsupervised feature selection by Maung and Schweitzer
  6. Multi-Prediction Deep Boltzmann Machines by Goodfellow, Mirza, Courville, and Bengio
  7. Memoized Online Variational Inference for Dirichlet Process Mixture Models by Hughes, and Sudderth
  8. Learning word embeddings efficiently with noise-contrastive estimation by Andriy , and Kavukcuoglu
  9. Learning Stochastic Feedforward Neural Networks by Tang and Salakhutdinov
  10. Distributed Representations of Words and Phrases and their Compositionality by Mikolov, Sutskever, Chen, Corrado, and Dean
  11. Correlated random features for fast semi-supervised learning by McWilliams, Balduzzi, and Buhmann
  12. Convex Two-Layer Modeling by Aslan, Cheng, Zhang, and Schuurmans, and
  13. Approximate inference in latent Gaussian-Markov models from continuous time observations by Cseke, Opper, and Sanguinetti

I suggest reading Zajac’s summaries before diving in.