slide news, faq, metrics names

slide/slide_index.md

@@ -161,8 +161,8 @@ The predefined process faults that should be detected and isolated are listed in
 | 10  |   $pf^S_{G_{V4}}$   | Valve 4 position sensor fault                                 |
 | 11  |   $pf^{CV}_{3.1}$   | Fault of the 3.1 controller CV path – closing                 |
 | 12  |   $pf^{CV}_{4.1}$   | Fault of the 4.1 controller CV path – closing                 |
-| 13  |   $pf^A_{{V3}}$   | Valve 3 blockage in the current position - lock               |
-| 14  |   $pf^A_{{V4}}$   | Valve 4 blockage in the current position - lock               |
+| 13  |   $pf^A_{V3}$   | Valve 3 blockage in the current position - lock               |
+| 14  |   $pf^A_{V4}$   | Valve 4 blockage in the current position - lock               |
 | 15  |   $pf^C_{SH3}$   | Steam leakage from the pipeline upstream of stage 3 superheater |
 | 16  |   $pf^C_{SH4}$   | Steam leakage from the pipeline upstream of stage 4 superheater |
 
@@ -199,15 +199,17 @@ Code for the submission and usage instructions can be found in the [github repos
 <a name="evaluation"></a>
 Participants will be evaluated on the following metrices:
 - Diagnosis of process faults:
-  - True detection rate (TDR) - The percentage of samples does the diagnosis system state that a fault is detected when there is a fault in the system.
   - False alarm rate (FAR) - The percentage of samples does the diagnosis system state that a fault is detected when there is no fault in the system. We use 1 - FAR as a metric.
+  - True detection rate (TDR) - The percentage of samples does the diagnosis system state that a fault is detected when there is a fault in the system.
+
   - <a href="#faults_isolation_accuracy">Fault isolation accuracy (FIA)</a> - The average probability given to the true diagnosis for all samples when a fault is correctly detected.
 - Diagnosis of cybernetic attacks:
-  - True detection rate (TDR) - The percentage of samples does the diagnosis system state that a cybernetic attack is detected when there is a cybernetic attack in the system.
   - False alarm rate (FAR) - The percentage of samples does the diagnosis system state that a cybernetic attack is detected when there is no cybernetic attack in the system. We use 1 - FAR as a metric.
+  - True detection rate (TDR) - The percentage of samples does the diagnosis system state that a cybernetic attack is detected when there is a cybernetic attack in the system.
+
   - <a href="#cybernetic_attacks_isolation_accuracy">Cybernetic attacks isolation accuracy (CIA)</a>
 
-The final score is a weighted sum of these metrics. 
+The final score is a weighted sum of these metrics.
 Each score is calculated independently for each given sample (simulation step). All scores are averaged over the whole evaluation dataset.
 
 Fault diagnosis and cybernetic attack detection will be evaluated separately. However, you can take into account that:
@@ -222,8 +224,6 @@ The algorithm should output:
 
 The above mentioned scores will be calculated based on these output signals.
 
-### Detection rate
-True and False Detection rate will be calculated for both, process fault detection signal and cybernetic attack detection signal.
 
 ### Faults isolation accuracy
 <a name="faults_isolation_accuracy"></a>
@@ -250,7 +250,7 @@ The TIR and (1 - FIR) should be maximized (closer to 1 is better).
 We compute the isolation accuracy score as harmonic mean of TIR and (1 - FIR), i.e.
 
 $$
-\text{Isolation accuracy score} = \frac{2 \cdot \text{TIR} \cdot (1 - \text{FIR})}{\text{TIR} + (1 - \text{FIR})}
+\text{CIA} = \frac{2 \cdot \text{TIR} \cdot (1 - \text{FIR})}{\text{TIR} + (1 - \text{FIR})}
 $$
 ---