Visual Perception with Deep Learning

Google Tech Talks
April, 9 2008

ABSTRACT

A long-term goal of Machine Learning research is to solve highy
complex "intelligent" tasks, such as visual perception auditory
perception, and language understanding. To reach that goal, the ML
community must solve two problems: the Deep Learning Problem, and the
Partition Function Problem.

There is considerable theoretical and empirical evidence that complex
tasks, such as invariant object recognition in vision, require "deep"
architectures, composed of multiple layers of trainable non-linear
modules. The Deep Learning Problem is related to the difficulty of
training such deep architectures.

Several methods have recently been proposed to train (or pre-train)
deep architectures in an unsupervised fashion. Each layer of the deep
architecture is composed of an encoder which computes a feature vector
from the input, and a decoder which reconstructs the input from the
features. A large number of such layers can be stacked and trained
sequentially, thereby learning a deep hierarchy of features with
increasing levels of abstraction. The training of each layer can be
seen as shaping an energy landscape with low valleys around the
training samples and high plateaus everywhere else. Forming these
high plateaus constitute the so-called Partition Function problem.

A particular class of methods for deep energy-based unsupervised
learning will be described that solves the Partition Function problem
by imposing sparsity constraints on the features. The method can learn
multiple levels of sparse and overcomplete representations of
data. When applied to natural image patches, the method produces
hierarchies of filters similar to those found in the mammalian visual
cortex.

An application to category-level object recognition with invariance to
pose and illumination will be described (with a live demo). Another
application to vision-based navigation for off-road mobile robots will
be described (with videos). The system autonomously learns to
discriminate obstacles from traversable areas at long range.

This is joint work with Y-Lan Boureau, Sumit Chopra, Raia Hadsell,
Fu-Jie Huang, Koray Kavakcuoglu, and Marc'Aurelio Ranzato.

Speaker: Yann Le Cun
Computational and Biological Learning Lab,
Courant Institute of Mathematical Sciences,
New York University.

Category:

People & Blogs

Tags:

• google
• techtalks
• techtalk
• engedu
• talk
• talks
• googletechtalks
• education

[adelle0001]

Good job! ang ganda ng video!

[srgrimm13156]

Does anyone else think the intro guy looks like a young Jeff Goldblum?? ;)

[davepamn]

The convolutional network filters using sparsity logistic function prevents the lowering of the energy of everything in the system for new inputs and reaches the objective of low energy minimums of new inputs while keeping the energy high for everything.

The long range vision provides a predictable encoding for the direction of the robot. The robot processes the image, applies filters, and predicts direction.

I want an iTouch apply with this long range vision software.

[MosesDj000]

Blasphemy!

[Devilboy668]

Good job I love this research

[dings1337]

Nice. I'm still viewing it, but it seems to answer the exact questions that came into my mind today.

I've just begun to try to get into this stuff like to do image recognition with restricted boltzmann machines and i was wondering, how it can still recognize e.g. a face, if the picture has undergone a perspective transformation or some rotation, because, then the whole geometry maps completely different to the first layer of neurons.

Hopefully after viewing this, I understand that a lot better.