Auto-EncodingVariationalBayes(VAE)ICLR2014Cite:7659IntroductionProblem:InferenceandlearningwithdirectedprobabilisticmodelsMethod:VAEProblemsetting:TargetVariationalBoundEvidenceLowerBOund(ELBO)VariationalBoundProofforEq.(1)VariationalBoundProofforEq.(3)ReparameterizationtrickApplication:VariationalAutoencoder(VAE)•Application:VariationalAutoencoder(VAE)Application:VariationalAutoencoder(VAE)ExperimentsExperimentsMoreresultsMoreresultsApplicationtoNLP•Generatingsentencesfromacontinuousspace•LearningDiscourse-levelDiversityforNeuralDialogModelsusingConditionalVariationalAutoencodersMorereading•https://spaces.ac.cn/archives/5253•AdversarialAutoencoders•WassersteinAuto-Encoders
