Robust generalization to new concepts has long remained a distinctive feature of human intelligence. However, recent progress in deep generative models has now led to neural architectures capable of synthesizing novel instances of unknown visual concepts from a single training example. Yet, a more precise comparison between these models and humans is not possible because existing performance metrics for generative models (i.e., FID, IS, likelihood) are not appropriate for the one-shot generation scenario. Here, we propose a new framework to evaluate one-shot generative models along two axes: sample recognizability vs. diversity (i.e., intra-class variability). Using this framework, we perform a systematic evaluation of representative one-shot generative models on the Omniglot handwritten dataset. We first show that GAN-like and VAE-like models fall on opposite ends of the diversity-recognizability space. Extensive analyses of the effect of key model parameters further revealed that spatial attention and context integration have a linear contribution to the diversity-recognizability trade-off. In contrast, disentanglement transports the model along a parabolic curve that could be used to maximize recognizability. Using the diversity-recognizability framework, we were able to identify models and parameters that closely approximate human data.

Robust generalization to new concepts has long remained a distinctive feature of human intelligence. However, recent progress in deep generative models has now led to neural architectures capable of synthesizing novel instances of unknown visual concepts from a single training example. Yet, a more precise comparison between these models and humans is not possible because existing performance metrics for generative models (i.e., FID, IS, likelihood) are not appropriate for the one-shot generation scenario. Here, we propose a new framework to evaluate one-shot generative models along two axes: sample recognizability vs. diversity (i.e., intra-class variability). Using this framework, we perform a systematic evaluation of representative one-shot generative models on the Omniglot handwritten dataset.

Our framework integrates a symbolic approach to represent causal law generation, with non-parametric Bayesian categorization to model latent categories, emphasizing the constructive nature of causal belief formation, in which both the content and extension of our causal concepts are generated rather than pre-specified. The constructive nature of the PCFG calls upon a potentially infinite set of possible causal functions, yet is governed by the preference for parsimony, and encourages systematic composition (see also Goodman et al., 2008; Bramley et al., 2018). The extended Dirichlet Process for category construction goes beyond a hierarchical Baysian modeling approach where categories are pre-defined as inductive biases (e.g. Griffiths and Tenenbaum, 2009; Goodman et al., 2011), and thus better captures the flexibility of human generalization behaviors (see also Kemp et al., 2010). This method draws a close link with probabilistic program induction models (e.g.