AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Analyzing the Variety Loss in the Context of Probabilistic Trajectory Prediction

Luca Anthony Thiede, Pratik Prabhanjan Brahma

ICCV 2019 · Probabilistic and scientific modeling › Molecules and proteins · approach: Multimodal trajectory prediction · 72 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
square and renormalise the learned PDF to undo MoN dilationMoN-trained toy generative model · Toy mixture of GaussiansJensen-Shannon divergence (lower is better): 0.1282 -> 0.0191Improved
apply compensation transformation to the marginalized PDFLSTM MoN generator · NGSIM time step 1marginalized log likelihood: -5.48 -> -5.28Improved
apply compensation transformation to the marginalized PDFLSTM MoN generator · NGSIM time step 2marginalized log likelihood: -5.46 -> -5.13Improved
apply compensation transformation to the marginalized PDFLSTM MoN generator · NGSIM time step 3marginalized log likelihood: -5.50 -> -5.12Improved
apply compensation transformation to the marginalized PDFSocial-GAN · Zara1 time step 1marginalized log likelihood: -5.57 -> -5.24Improved
apply compensation transformation to the marginalized PDFSocial-GAN · Zara1 time step 2marginalized log likelihood: -5.87 -> -5.77Improved
apply compensation at later time steps t>=3 and harder pedestrian dataSocial-GAN · Zara1 time step 3+marginalized log likelihoodHurt
start MoN training with low N and slowly increase to 128MoN-trained toy generative model · Toy mixture of Gaussianslearned PDF qualityImproved
increase N in the MoN lossMoN loss, 1-D PDF family · Toy PDF family exponent searchexponent k minimising MoN: k approaches 0.5 as N growsImproved
use 10-dimensional PDF instead of 1-DMoN loss · Toy PDF family exponent searchexponent k minimising MoN: k close to 0.5 even for small NMixed

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