13. 多層ネットワークの階層構造の例
• 特定の物体だけに選択的に反応するユニット
Building High-level Features Using Large Scale Unsupervised Learning
引用:http://static.googleusercontent.com/media/research.google.com/ja//
pubs/archive/38115.pdf
100,000
et, only
d the lo-
ned the
uracy of
ith pre-
contrast
for face
be seen,
n learns
tractors.
age, the
reshold,
an input
than 0.
s (blue).
between
on tech-
approach can be susceptible to local minima. Results,
shown in Figure 13, confirm that the tested neuron
indeed learns the concept of faces.
Figure 3. Top: Top 48 stimuli of the best neuron from the
test set. Bottom: The optimal stimulus according to nu-
merical constraint optimization.
4.5. Invariance properties
We would like to assess the robustness of the face de-
tector against common object transformations, e.g.,
translation, scaling and out-of-plane rotation. First,
Building high-level features using large-scale unsupervised learning
e (left) and out-of-plane (3D) rotation (right)
perties of the best feature.
Figure 6. Visualization of the cat face neu
human body neuron (right).
For the ease of interpretation, these da
positive-to-negative ratio identical to the