Travelled to:1 × France
1 × Israel
1 × United Kingdom
2 × USA
Collaborated with:I.Sutskever ∅ G.E.Hinton R.B.Grosse K.Swersky G.E.Dahl
Talks about:learn (4) network (3) hessian (3) neural (3) optim (3) curvatur (2) recurr (2) free (2) deep (2) momentum (1)
Person: James Martens
DBLP: Martens:James
Contributed to:
20152013201220112010Wrote 7 papers:
- ICML-2015-MartensG #approximate #network #optimisation
- Optimizing Neural Networks with Kronecker-factored Approximate Curvature (JM, RBG), pp. 2408–2417.
- ICML-c3-2013-SutskeverMDH #learning #on the
- On the importance of initialization and momentum in deep learning (IS, JM, GED, GEH), pp. 1139–1147.
- ICML-2012-MartensSS
- Estimating the Hessian by Back-propagating Curvature (JM, IS, KS), p. 126.
- ICML-2011-MartensS #learning #network #optimisation
- Learning Recurrent Neural Networks with Hessian-Free Optimization (JM, IS), pp. 1033–1040.
- ICML-2011-SutskeverMH #generative #network
- Generating Text with Recurrent Neural Networks (IS, JM, GEH), pp. 1017–1024.
- ICML-2010-Martens #learning #optimisation
- Deep learning via Hessian-free optimization (JM), pp. 735–742.
- ICML-2010-Martens10a #learning #linear
- Learning the Linear Dynamical System with ASOS (JM), pp. 743–750.