Strike (with) a Pose:
Neural Networks Are Easily Fooled by Strange Poses of Familiar Objects
Michael A. Alcorn, Qi Li, Zhitao Gong, Chengfei Wang, Long Mai, Wei-Shinn Ku, and Anh Nguyen
Motivation
2/11
Questions you will be able to answer after this talk
1.
2.
3.
3/11
How do artificial neural networks work?
input = observed “features”/things you measured
output = “label”, i.e., the thing you’re trying to predict
nodes/neurons = computational unit
weights/connections (wis) = what the neural network is trying to learn
Training Algorithm
4/11
How are artificial neural networks
trained for computer vision tasks?
Step #1: Acquire labeled dataset
Step #2: Train a Convolutional Neural Network
5/11
Do artificial neural networks understand
what they are looking at?
3D School Bus
3D School Bus
Real School Bus
6/11
Do artificial neural networks understand
what they are looking at?
#1: Collected 30 3D objects found in a traffic environment
7/11
Do artificial neural networks understand
what they are looking at?
#2: Randomly sampled object poses and recorded neural network predictions
Median Misclassification Rate = 97%!
8/11
Do artificial neural networks understand
what they are looking at?
#3: Altered individual parameters of correctly recognized poses
9/11
Do artificial neural networks understand
what they are looking at?
#4: Used black box optimization to discover specific failures
10/11
Do artificial neural networks understand
what they are looking at?
Links
11/11