2 Methods 2.1 VAEsasaConstrainedOptimisationProblem VAEsmodelthedistributionofi.i.d.dataD ={xi}Ni=1 asthemarginal Y

Concurrently, we introduce the associated optimisation algorithm, which is inspired by GECO [25]--the latter does not always lead to good encodings (e.g., Sec.