(for MINERvA experiment)
src: nobelprize.org
Main INjector ExpeRiment \(\nu\)-A
by changing distance between horns one can change energy spectrum
by changing horns polarization one can switch between neutrino and anti-neutrino mode
LE event example
courtesy of G. Perdue
courtesy of G. Perdue
tracking based algorithms fail for high energy events
"by eye" method is very often more accurate
idea: use algorithms for images analysis and pattern recognition
ImageNet is an image database
Siberian Husky or Alaskan Malamute?
If you can't explain it simply, you don't understand it well enough.
Albert Einstein
epoch = one loop over the whole training sample
for each feature vector weights are updated using gradient descent method
target: \(y = 0, 1\)
not really efficient for classification
imagine having some data ~ 100
Logistic function: \[g(z) = \frac{1}{1 + e^{-z}}\]
We can do classification
We can do regression
But real problems are nonlinear
Feature vector: \[(x,y) \rightarrow (x,y,x^2,y^2)\]
Hypothesis: \[h (x) = \frac{1}{1 + e^{-w_0 - w_1x - w_2y - w_3x^2 - w_4y^2}}\]
In general, adding extra dimension by hand would be hard / impossible. Neural networks do that for us.
\(x_1\) | 0 | 1 | 0 | 1 |
\(x_2\) | 0 | 0 | 1 | 1 |
AND | 0 | 0 | 0 | 1 |
\[h(x) = \frac{1}{1 + e^{-w^Tx}}\]
Intuition:
x XOR y = (x AND NOT y) OR (y AND NOT x)
Hidden neuron #1:
0, 0 = 0.000555
0, 1 = 0.000001
1, 0 = 0.263002
1, 1 = 0.000827
Hidden neuron #2:
0, 0 = 0.000567
0, 1 = 0.290434
1, 0 = 0.000002
1, 1 = 0.001137
Final results:
0 XOR 0 = 0.035760
0 XOR 1 = 0.956746
1 XOR 0 = 0.956866
1 XOR 1 = 0.026566
The more complicated problem is the more neurons we need
src: deeplearning.net
src: wildml.com
src: arxiv
src: wildml.com
Machine Learning for MINERvA Physics Reconstruction
the first goal is to use CNN to find the primary vertex in nuclear target region
each event is represented by 3 "pictures" - different views at the detector
event examples courtesy of G. Perdue
Convolution layer | No. of filters | Filter size | Pool size |
1 | 12 | (8,3) | (2,1) |
2 | 20 | (7,3) | (2,1) |
3 | 28 | (6,3) | (2,1) |
4 | 36 | (6,3) | (2,1) |
In order to attain the impossible, one must attempt the absurd.
Miguel de Cervante
Target | Track-based score [%] | CNN-based score [%] | Improvement [%] |
---|---|---|---|
1 | 89.4 | 95.7 | 6.3 |
2 | 85.8 | 96.0 | 10.2 |
3 | 84.0 | 94.6 | 10.6 |
4 | 84.1 | 92.6 | 8.5 |
5 | 86.9 | 94.6 | 7.7 |
ML approach outperforms track-based reconstruction
It improves efficiency and purity
And this is just the beginning:
CNN would fail to access Physical Review database