Vision Day Schedule Time Speaker and Collaborators Affiliation Title a General Preprocessing Method for Improved Performance of Epipolar Geometry Estimation Algorithms on the Expressive Power of Deep Learning: a Tensor Analysis

Visual Learning of Arithmetic Operations Yedid Hoshen and Shmuel Peleg ­ HUJI A simple Neural Network model is presented for end­to­end visual learning of arithmetic operations from pictures of numbers. The input consists of two pictures, each showing a 7­digit number. The output, also a picture, displays the number showing the result of an arithmetic operation (e.g., addition or subtraction) on the two input numbers. The concepts of a number, or of an operator, are not explicitly introduced. This indicates that addition is a simple cognitive task, which can be learned visually using a very small number of neurons. Other operations, e.g., multiplication, were not learnable using this architecture. Some tasks were not learnable end­to­end (e.g., addition with Roman numerals), but were easily learnable once broken into two separate sub­tasks: a perceptual Character Recognition and cognitive Arithmetic sub­tasks. This indicates that while some tasks may be easily learnable end­to­end, other may need to be broken into sub­tasks.