Designing Perceptual Puzzles by Differentiating Probabilistic Programs

04/26/2022
by   Kartik Chandra, et al.
4

We design new visual illusions by finding "adversarial examples" for principled models of human perception – specifically, for probabilistic models, which treat vision as Bayesian inference. To perform this search efficiently, we design a differentiable probabilistic programming language, whose API exposes MCMC inference as a first-class differentiable function. We demonstrate our method by automatically creating illusions for three features of human vision: color constancy, size constancy, and face perception.

READ FULL TEXT

page 5

page 6

03/06/2019

LF-PPL: A Low-Level First Order Probabilistic Programming Language for Non-Differentiable Models

We develop a new Low-level, First-order Probabilistic Programming Langua...
08/06/2019

Functional probabilistic programming for scalable Bayesian modelling

Bayesian inference involves the specification of a statistical model by ...
04/08/2021

Sound Probabilistic Inference via Guide Types

Probabilistic programming languages aim to describe and automate Bayesia...
10/30/2019

Bayesian causal inference via probabilistic program synthesis

Causal inference can be formalized as Bayesian inference that combines a...
08/20/2019

Reactive Probabilistic Programming

Synchronous reactive languages were introduced for designing and impleme...
07/20/2020

Automating Involutive MCMC using Probabilistic and Differentiable Programming

Involutive MCMC is a unifying mathematical construction for MCMC kernels...
08/24/2016

Towards Bayesian Deep Learning: A Framework and Some Existing Methods

While perception tasks such as visual object recognition and text unders...