AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Neural approximation of Wasserstein distance via a universal architecture for symmetric and factorwise group invariant functions

Samantha Chen, Yusu Wang

NeurIPS 2023 · Learning theory and algorithms › Prediction and decision losses · approach: Theory of neural network sensitivity · 0 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 ProductNet instead of WPCE (Siamese autoencoder with Sinkhorn loss)1-Wasserstein approximation · noisy-sphere-3, input size [100, 300]mean relative error: 0.341 ± 0.202 -> 0.046 ± 0.043Improved
Evaluate on input sizes unseen at training time vs WPCEProductNet · ModelNet-small, input size [300, 500]mean relative error: 0.241 ± 0.198 -> 0.111 ± 0.086Improved
Use ProductNet instead of WPCE on seen sizes1-Wasserstein approximation · ModelNet-small, input size [20, 200]mean relative error: 0.077 ± 0.075 -> 0.084 ± 0.077Hurt
Use ProductNet instead of Siamese DeepSets1-Wasserstein approximation · uniform, input size [200, 300]mean relative error: 0.917 ± 0.869 -> 0.131 ± 0.096Improved
Replace WPCE with ProductNet to cut training costProductNet · RNAseq(2k)training time: 14hr 26min -> 15minImproved
Replace WPCE with ProductNet to cut training costProductNet · noisy-sphere-3epochs to convergence: 100 -> 20Improved
Compare ProductNet with Sinkhorn (eps=0.1) on high-dimensional unseen-size dataProductNet · RNAseq, input size [300, 500]mean relative error: 0.048 ± 0.006 -> 0.292 ± 0.041Hurt
Use ProductNet instead of Sinkhorn (eps=0.1) on inference speedProductNet · ModelNet-largetime for 500 computations (s): 239.448 -> 1.174Improved
Use ProductNet instead of Sinkhorn (eps=0.1) on accuracyProductNet · noisy-sphere-6, input size [100, 300]mean relative error: 0.137 ± 0.122 -> 0.015 ± 0.014Improved
Lower Sinkhorn regularization from eps=0.1 to eps=0.01Sinkhorn · ModelNet-smallmean relative error and time (s): 0.187 ± 0.232 -> 0.011 ± 0.003; 2.074 -> 104.712Mixed
Lower Sinkhorn regularization from eps=0.1 to eps=0.01Sinkhorn · noisy-sphere-6mean relative error: 0.137 ± 0.122 -> 0.326 ± 0.135Hurt
Use PointNet (max aggregator) instead of DeepSets (sum) as WPCE encoder (ablation)WPCE · 1-Wasserstein approximationapproximation qualityNo clear effect

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