Please enable JavaScript.

Coggle requires JavaScript to display documents.

FBI - MISC - Coggle Diagram

- - - - Aim to identify categories based on features
  - - - unsupervised is harder
        
        requires more and better data
    - - During training process to optimise
      - 5 fold is standard
        
        Split into 5. Train on 4/5, test on 1/5. Rotate. Retrain on all data then do final testing
    - - Discard lowest weighted features after each round
  - - - On data set aside
  - - - Risks overfitting
    - - Allows for projection
        
        Projection vector = separation line
        
        Distance to decision boundary from origin
        
        Allows for feature space to be rotated -> decision boundary becomes new axis
        
        Class a < p < class b
  - - - Confound corrected z-score
- - - - Beyond chance overlap of task conenctivity and sleep connectivity
        
        task specific and cross-frequency
    - - Even when there's not a person in video
      - via mirror neurons
  - - - less remembering
- - - - . . . .
      - Low freq = 1 Hz = inhibitory
      - High freq = >5Hz (100Hz doable) = Exc
    - - ,,, ,,, ,,,
      - More accurate mimicking of brain patterns?
      - Continuous is inhibitory. Intermittent is excitatory
    - - Smaller prepulse
      - Used with PD
        
        PD shows lack of inhibition. Meds help to normalise
      - <6ms is inhibitory
      - 10 seconds is facilitative
    - - Used to mediate dopamine levels in the substantia nigra
        
        Downstream effects
        
        Targets networks, not just areas
    - - Probing. Diagnostic. Interference with ongoing processes
- - - - Restructuring effect