Travelled to:1 × France
1 × USA
2 × China
Collaborated with:M.R.DeWeese S.Ganguli M.Mudigonda B.Poole P.Battaglino E.A.Weiss N.Maheswaranathan
Talks about:learn (2) nonequilibrium (1) thermodynam (1) hamiltonian (1) unsupervis (1) stochast (1) gradient (1) without (1) probabl (1) minimum (1)
Person: Jascha Sohl-Dickstein
DBLP: Sohl-Dickstein:Jascha
Contributed to:
2015201420142011Wrote 4 papers:
- ICML-2015-Sohl-DicksteinW #learning #using
- Deep Unsupervised Learning using Nonequilibrium Thermodynamics (JSD, EAW, NM, SG), pp. 2256–2265.
- ICML-c1-2014-Sohl-DicksteinMD #monte carlo
- Hamiltonian Monte Carlo Without Detailed Balance (JSD, MM, MRD), pp. 719–726.
- ICML-c2-2014-Sohl-DicksteinPG #optimisation #performance #probability #scalability
- Fast large-scale optimization by unifying stochastic gradient and quasi-Newton methods (JSD, BP, SG), pp. 604–612.
- ICML-2011-Sohl-DicksteinBD #learning #probability
- Minimum Probability Flow Learning (JSD, PB, MRD), pp. 905–912.