A neural approach to the analysis of CHIMERA experimental data

1
A neural approach to the analysis of CHIMERA
experimental data

• CHIMERA Collaboration
• S.Aiello1, M. Alderighi2,3, A.Anzalone4,
  M.Bartolucci5, G.Cardella1, S.Cavallaro4,7, M.
  DAgostino6 ,E.DeFilippo1, E.Geraci4, M.Geraci1,
  F.Giustolisi4,7, P.Guazzoni3,5, M.Iacono Manno4,
  G.Lanzalone1,7, G.Lanzanò1, S.LoNigro1,7,
  G.Manfredi5, A.Pagano1, M.Papa1, S.Pirrone1,
  G.Politi1,7, F.Porto4,7, S.Russo5,
  S.Sambataro1,7, G.Sechi2,3, L.Sperduto4,7,
  C.Sutera1, L.Zetta3,5
• 1Istituto Nazionale di Fisica Nucleare, sez di
  Catania, Catania, Italy
• 2Istituto di Fisica Cosmica, CNR, Milano, Italy
• 3Istituto di Fisica Nucleare, sez. di Milano,
  Milano, Italy
• 4 Istituto di Fisica Nucleare, Laboratorio
  Nazionale del Sud, Catania, Italy
• 5Dipartimento di Fisica dellUniversita, Milano,
  Italy
• 6Dipartimento di Fisica dell'Universita degli
  Studi and Istituto di Fisica Nucleare, sez. di
  Bologna, Bologna,Italy
• 7Dipartimento di Fisica dell'Universita,
  Catania, Italy

2
Outline

• Detector characteristics
• Automatic data analysis
• Proposed approaches
• Our neural approach
• System overview
• Results

3
CHIMERA (Charged Heavy Ion Mass and Energy
Resolving Array)
9 wheels
1192 Si-CsI(TI) detection cells
4
Detection cell
5
Scatter plot from CHIMERA
58 Ni 27 Al Einc 30 AMev

• sparse data
• low S/N
• density variation
• high frequency noise
• characteristic frequency ridges/valleys
• low frequency background

6
banana extraction
?
D E-Si
Fast-CsI(TI)
Counts
D E-Si
7
1-D frequency distribution
Z-lines
D E-Si
Fast-CsI(TI)
Counts
Fast-CsI(TI)
D E-Si
D E-Si
8
Proposed approach

• FFT not satisfactory results
• filtering edge detection ill-posed problem
• contextual image segmentation Benkirane et al.
  95 Canny filtering a priori information
  not easily applicable
• interactive technique unpractical for a lot
  of spectra

yet, density modulation can be easily perceived
by sight
9
Our solution

• Using emergent perception mechanisms of
  biological visual systems
• Grossbergs neural networks
• mathematically defined
• extract information from the global structure of
  data (rather local relationship)
• no training
• successfully applied to SAR and satellite images
  (noisy and incomplete)

10
Implementation

• 2 levels of neural networks for cluster
  determination
• Procedural algorithms for frequency distribution
  construction
• Matlab (PC Pentium II, 400MHz)
• 500 ? 500 pixel processing windows

11
Neural system
Level 2 oriented completion (Bipole Filter)
BF net
Level 1 Adaptive Density Discrimination
ADD net
Window
Input
12
Level 1 ADD net

• on-center off-surround shunting network
• density information processing
• comparison between on-center and off-surround
  areas
• low-pass filtering of the spatial frequencies in
  the input window sensitivity to ridge-valley
  modulation
• clusters as incomplete and irregular strips

13
ADD net
14
Level 2 BF nets

• additive networks
• long-term cooperation along selected directions
• bipole filters
• different filtering masks according to hyperbolic
  trends of data
• clusters as complete strips

Ex.
105
135
15
Example 1
valley clusters
16
Example 2
ridge clusters
17
Conclusions

• Grossbergs approach is good for automatic
  determination of bananas
• Density processing is
• dependent on the image structure only
• independent from the underlying physics
• Intensive computation (500 ? 500 neurons)
• Processing whole matrices and improving algorithm
  efficiency as future works