AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

A Semantic Loss Function for Deep Learning with Symbolic Knowledge

Jingyi Xu, Zilu Zhang, Tal Friedman, Yitao Liang, Guy Broeck

ICML 2018 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 70 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
Add exactly-one semantic loss to an MLPMNIST semi-supervised classification with 100 labels · MNIST, 100 labelsaccuracy (%): 78.46 ($\pm$1.94) -> 98.38 ($\pm$0.51)Improved
Add exactly-one semantic loss to an MLPMNIST semi-supervised classification with 1000 labels · MNIST, 1000 labelsaccuracy (%): 94.26 ($\pm$0.31) -> 98.78 ($\pm$0.17)Improved
Add exactly-one semantic loss to an MLPMNIST fully supervised classification · MNIST, all labelsaccuracy (%): 99.34 ($\pm 0.08$) -> 99.36 ($\pm$0.02)No clear effect
Add exactly-one semantic loss to an MLPFASHION semi-supervised classification · FASHION, 100 labelsaccuracy (%): 69.45 ($\pm 2.03$) -> 86.74 ($\pm 0.71)Improved
Add exactly-one semantic loss to an MLPFASHION semi-supervised classification · FASHION, 500 labelsaccuracy (%): 78.12 ($\pm 1.41) -> 89.49 ($\pm 0.24)Improved
Add exactly-one semantic loss to an MLPFASHION fully supervised classification · FASHION, all labelsaccuracy (%): 89.87 -> 89.81No clear effect
Add exactly-one semantic loss to a CNNCIFAR-10 semi-supervised classification · CIFAR-10, 4000 labelsaccuracy (%): 77.13 -> 81.79Improved
Add exactly-one semantic loss to a CNNCIFAR-10 fully supervised classification · CIFAR-10, all labelsaccuracy (%): 90.96 -> 90.92No clear effect
Add semantic loss for valid simple paths5-layer MLP on grid shortest paths · Grid shortest pathcoherent accuracy (%): 5.62 -> 28.51Improved
Add semantic loss for valid simple paths5-layer MLP on grid shortest paths · Grid shortest pathincoherent accuracy (%): 85.91 -> 83.14Hurt
Add semantic loss for valid simple paths5-layer MLP on grid shortest paths · Grid shortest pathconstraint accuracy (%): 6.99 -> 69.89Improved
Add semantic loss for valid total orderings3-layer MLP on sushi preference prediction · Sushi preference predictioncoherent accuracy (%): 1.01 -> 13.59Improved

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