Timezone: »
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.
Author Information
Victor Boutin (Brown university)
Lakshya Singhal (Indian Institute of Technology Delhi)
Xavier Thomas (Manipal Institute of Technology)
Thomas Serre (Brown University)
More from the Same Authors
-
2022 : The emergence of visual simulation in task-optimized recurrent neural networks »
Alekh Karkada Ashok · Lakshmi Narasimhan Govindarajan · Drew Linsley · David Sheinberg · Thomas Serre -
2022 Poster: Meta-Reinforcement Learning with Self-Modifying Networks »
Mathieu Chalvidal · Thomas Serre · Rufin VanRullen -
2022 Poster: A Benchmark for Compositional Visual Reasoning »
Aimen Zerroug · Mohit Vaishnav · Julien Colin · Sebastian Musslick · Thomas Serre -
2022 Poster: Harmonizing the object recognition strategies of deep neural networks with humans »
Thomas FEL · Ivan F Rodriguez Rodriguez · Drew Linsley · Thomas Serre -
2022 Poster: What I Cannot Predict, I Do Not Understand: A Human-Centered Evaluation Framework for Explainability Methods »
Julien Colin · Thomas FEL · Remi Cadene · Thomas Serre -
2021 Poster: Tracking Without Re-recognition in Humans and Machines »
Drew Linsley · Girik Malik · Junkyung Kim · Lakshmi Narasimhan Govindarajan · Ennio Mingolla · Thomas Serre -
2021 Poster: Look at the Variance! Efficient Black-box Explanations with Sobol-based Sensitivity Analysis »
Thomas FEL · Remi Cadene · Mathieu Chalvidal · Matthieu Cord · David Vigouroux · Thomas Serre -
2020 Poster: Stable and expressive recurrent vision models »
Drew Linsley · Alekh Karkada Ashok · Lakshmi Narasimhan Govindarajan · Rex Liu · Thomas Serre -
2020 Spotlight: Stable and expressive recurrent vision models »
Drew Linsley · Alekh Karkada Ashok · Lakshmi Narasimhan Govindarajan · Rex Liu · Thomas Serre -
2020 Session: Orals & Spotlights Track 29: Neuroscience »
Aasa Feragen · Thomas Serre -
2018 Poster: Learning long-range spatial dependencies with horizontal gated recurrent units »
Drew Linsley · Junkyung Kim · Vijay Veerabadran · Charles Windolf · Thomas Serre -
2016 Poster: How Deep is the Feature Analysis underlying Rapid Visual Categorization? »
Sven Eberhardt · Jonah G Cader · Thomas Serre -
2013 Poster: Neural representation of action sequences: how far can a simple snippet-matching model take us? »
Cheston Tan · Jedediah M Singer · Thomas Serre · David Sheinberg · Tomaso Poggio