AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Relative gradient optimization of the Jacobian term in unsupervised deep learning

Luigi Gresele, Giancarlo Fissore, Adrián Javaloy, Bernhard Schölkopf, Aapo Hyvarinen

NeurIPS 2020 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Flow-based density estimation · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use relative-gradient updates instead of ordinary gradients to optimize the Jacobian termFully connected invertible neural networks · Gradient computation experiment; data dimensions from 10 to beyond 20,000gradient computation time: of the order of 100Improved
Use relative-gradient updates instead of ordinary gradientsTwo-layer neural network trained on 1,000 datapoints of dimensionality 2 · Full-batch optimizationlossNo clear effect
Use the proposed relative-gradient density estimator instead of MADE MoGUnconditional density estimation · POWER testlog-likelihood: 0.375\pm 0.013 -> 0.065\pm0.013Hurt
Use the proposed relative-gradient density estimator instead of Real NVP (10)Unconditional density estimation · GAS testlog-likelihood: 8.357\pm 0.019 -> 6.978\pm0.020Hurt
Use the proposed relative-gradient density estimator instead of MADE MoGUnconditional density estimation · HEPMASS testlog-likelihood: -18.368\pm 0.019 -> -21.958\pm0.019Hurt
Use the proposed relative-gradient density estimator instead of Real NVP (10)Unconditional density estimation · MINIBOONE testlog-likelihood: -11.795\pm 0.453 -> -13.372\pm 0.450Hurt
Use the proposed relative-gradient density estimator instead of Real NVP (10)Unconditional density estimation · BSDS300 testlog-likelihood: 153.28\pm 1.78 -> 151.12\pm0.28Hurt
Use the proposed relative-gradient density estimator instead of MADE MoGUnconditional density estimation · MNIST testlog-likelihood: -1038.5 \pm 1.8 -> -1375.2 \pm 1.4Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy