BayesiaLab 4.5: new features

• Data
  • Database saved with graph
  • Evidence scenario file
  • Default missing and filtered values lists
  • Associated continuous variables as targets for discretization and aggregation
  • Optional database loading
  • Import/Export forbidden arcs in the menus
  • KMeans discretization algorithm upgrade
  • States modification without loosing the database association
• Network
  • Filtered states
  • Excluded nodes
  • Global editor and indicator of temporal indices
  • Cost indicator
  • CPT of the selected nodes in the graph report
  • List of the excluded nodes in the graph report
  • Forbidden arcs sorted in the graph report
  • Structural coefficient for each network
• Learning
  • Temporal indices taken into account in the spanning tree
  • Last used learning kept and saved
  • Saving continuous values in the mutliple clustering's databases
  • Cluster with sorted numerical states
  • Modification of state values computation in the clustering
• Inference
  • Improvement of the junction tree's performance
  • Improvement of the junction tree's creation
  • Improvement of the complexity reducer
  • Observation of fixed probabilities
  • Evidence scenario file for batch exploitation
  • Evidence scenario file for interactive inference and updating
  • Evidence scenario file for temporal networks
• Analysis
  • Initial probability displayed in the target sensitivity analysis
  • Unification of evidence contexts
  • Comments and long names displayed in the parameter sensitivity analysis
  • KL taking into account the filtered values
  • KL and global contribution displayed in the arc comments
  • Choice of the independence test G or Chi²
  • G_KL-test on the network for the relationship analysis
  • Independence test on the network or the database for the relationship analysis, total effects and target analysis
  • Index of continuous node's states displayed int the reports
  • Tooltip in the dendrogram of the variable clustering
  • Degree of freedom displayed for the occurrence matrix an the mosaic
  • Mosaics copied in HTML format
  • Mosaic can be computed from network or from database
• Monitors
  • Evidences saved in the evidence scenario file
  • Color of the nodes and monitor's bars according to the kind of evidences
  • Changing target node and target state
  • Centering nodes corresponding to the selected monitors
• Interface
  • Optimization of multiple deletion with a database
  • Selection preserved when changing the mode
  • Ability to hide the node names
  • Working directories
  • Displaying client identifier
  • New node selections
  • New arc selections
  • Compatibility improvements with Mac OS X Leopard
  • New keyboard shortcuts
• Tools
  • New Tools menu
  • Graphs comparison
  • Arc confidence analysis
  • Targeted cross validation
  • Structure graphical comparator
  • Network extractor
• Settings
  • Text fields replaced by spinners
  • Reorganization of database settings
  • Comment font
  • Choice of the decimal separator for the export of the numbers
  • Option to write or not BOM for UTF
  • Settings of the various directories
  • Minimum size of the clusters for the clustering
• Security
  • BayesiaLicenseServer version 3.0.X
  • Network locking
• Internationalizing
  • Chinese locale
  • Spanish locale

Data

1. Hard evidence on a state of a node
2. Likelihood distribution over the states of a node (noted l{...} )
3. Probability distribution over the states of a node (fixed distribution with exact inference, computation of the corresponding likelihoods with approximate inference) (noted p{...} )

0;?Valve1?:OK;?Valve2?:OK;?Valve3?:OK //All the valves are working
2;?Valve1 t+1?:l{OK:0.8;RC:0.9;RO:0.9}
20;?Valve2 t+1?:l{OK:0.3;RO:0.3;RC:0.3};?Valve1 t+1?:p{OK:0.2;RO:0.4;RC:0.4}
30;?Valve3 t+1?:OK;?Valve1 t+1?:p{OK:0;RO:0.8;RC:0.2}

?Smoker?:Yes;?Age?:25.5;?Bronchitis?:p{Yes:0.8;No:0.2} //Young smoker with a large probability of bronchitis
?Smoker?:No;?Age?:70;Dyspnea:l{Yes:0.8;No:0.5} //Non-smoker senior person 

1. When one of the created intervals size is below a specified threshold, the points contained in the interval are automatically merged with its neighbours.
2. A filter whose purpose is reducing outlier impact on discretization has been introduced. The filter value allows computing the width of the sliding window which will balance each value of the database. The size of this window is the given percentage of the distinct values number. In order to activate this filter, the filter value must respect two conditions: first the filter value must be superior or equal to 2, second the number of distinct numeric values in the database must be superior to (2 + width of window).  

Network

• if the value of the variable Test is "True" then the value of the varaible Analysis can be "Positive" or "Negative" according to a certain probability distribution
• if the value of the variable Test is "False" then the value of the variable Analysis will be the filtered value "Not applicable" (or "*")

1. No initially filtered state in the continuous node, filtered values in the database: an interval of width 1E-7 will be added after the intervals defined in the node.
2. Initially filtered state in the continuous node and no filtered value in the database: the imported values which correspond to the interval of the filtered state are treated as filtered values.
3. Initially filtered state in the continuous node and filtered values in the database: the imported values which correspond to the interval of the filtered state are treated as filtered values and the imported filtered values are also associated with this interval.

Learning

Inference

• by maintaining the left mouse button pressed while choosing the desired probability level, or
• directly by editing the probability value thanks to a double-click on the value.
• A click on the name of the state (on the right) fix the current probability value (the probability bar is green).

1. Simply setting the probabilities: When the probabilities are valildated with the light green button, the likelihoods associated to the states of the node are computed again in order to make the marginal probability distribution correspond to the distribution entered by the user. It is, in fact, an indirect capture of the likelihoods. You must note that, at the next observation of another node, the probability distribution of this node will change because the likelihoods are not computed again.
   The result will be displayed with light green bars as the likelihoods:
   
   The observed node takes the light green color of the evidence.
2. Fixing the probabilities: When the probabilities are valildated with the mauve button, the likelihoods associated to the states of the node are computed as in the previous case. However, after each new evidence entered on the other nodes, the likelihoods are computed again in order to make the marginal probability distribution correspond (if the solution exists) to the distribution initially entered by the user. Fixing probabilities is also done in the evidence scenario files with the notation p{...}. You must note that fixing probabilities is only valid for the exact inference. If the approximate inference is used, fixing probabilities is considered like simply setting the probabilities.
   The result will be displayed with mauve bars:
   
   The observed node takes the mauve color of the evidence.

• Labeling
• Inference
• Most probable explanation labeling
• Most probable explanation inference
• Joint probability

1. Hard evidence on a state of a node
2. Likelihood distribution over the states of a node
3. Probability distribution over the states of a node (fixed distribution with exact inference, computation of the corresponding likelihoods with approximate inference)

Analysis

1. The hard evidence on a node's state is noted:
   <Node>: <State>
2. The observation of a likelihood distribution on the states of a node is noted:
   <Node>: l{ <State1>: <Likelihood1 %> , ... , <StateN>: <LikelihoodN %>}
3. The observation of a probability distribution on a node (fixed distribution in exact inference, computation of the corresponding likelihoods in approximate inférence) is noted:
   <Node>: p{ <State1>: <Probability1 %> , ... , <StateN>: <ProbabilityN %>}

• Network: the Structure equivalent example number allows simulating a set of data to compute the standardized Pearson's residual. This number is, by default, the number of examples used for the last learning.
• Database: the standardized Pearson's residual is directly computed from the population of the associated database.

Monitors

1. Hard evidence on a state of a node
2. Likelihood distribution over the states of a node
3. Probability distribution over the states of a node (fixed distribution with exact inference, computation of the corresponding likelihoods with approximate inference)

• Green: hard evidence on a node
• Light green: observation of the likelihoods (direct or via the probability setting)
• Mauve: fixed probability observation

Interface

• all nodes,
• excluded nodes.

• all arcs,
• fixed arcs,
• temporal arcs,
• not-oriented arcs.

• Shift + M (validation mode): Mosaic analysis
• Shift + P (validation mode): Target dynamic profile
• P (in the structure comparator): Automatic layouting
• Q (validation mode): Adaptive questionnaire
• F (validation mode): Arc force analysis
• G (validation mode): Pearson's correlation
• H (validation mode): Node force analysis
• S (validation mode): Variable clustering
• C + click on a monitor (validation mode): Search of the corresponding node
• S + several selected monitors (validation mode): Selection of the corresponding nodes and fit of the display into the window if necessary
• Alt + clic on a state in a monitor (validation mode): Corresponding node and state considered as target

Tools

• Common arcs: same arcs in networks
• Inverted arcs: arcs which orientation changes in the comparison network
• Added arcs: arcs that exist only in the comparison network
• Deleted arcs: arcs that exist only in the reference network
• Common edges: same edges in networks
• Added edges: edges that exist only in the comparison network
• Deleted edges:  edges that exist only in the reference network
• Common V-Structures: same V-Structures in networks
• Added V-Structures: V-Structures that exist only in the comparison network
• Deleted V-Structures:  V-Structures that exists only in the reference network

1. The learning context: Reminds the learning method and the number of data samples. It also indicates the structural complexity coefficient used (it is the same as the one used in the initial network).
2. Arc confidence analysis: Lists of arcs grouped by three colored types
   
   • Black: the arcs that exist both in the reference structure and the networks from the samples.
     
     • Arc frequency represent how often the arc appeared with the same orientation as in the sample networks.
     • Inverted arc frequency represent how often the arc appeared with the reverse orientation as in the sample networks.
     • Edges frequency represent how often the arc appeared without any orientation as in the sample networks (the equivalence class of the learnt networks is used) .
     • Total frequency is the sum of all previous ones. It indicates the overall strenght between the variables.
   • Blue: the arcs that exist in at least one sample network but that do not exist in the reference network. In this case, the frequencies are displayed with a negative value. The reference orientation of arc is arbitrarily given by the first arc found in the first semple network.
     
     • Arcs frequency represent how often this arc appeared with the same orientation as the first.
     • Inverted arc frequency represent how often this arc appeared with the reverse orientation compared with the first.
     • Edge frequency represent how often the arc appeared without any orientation (equivalence class).
     • Total frequency is the sum of all previous ones, it indicates the overall strength of a relationship that does not exist in the reference network.
   • Red: arcs that exist in the reference structure but that have never been found in any learnt sample.
3. V-Structures confidence analysis: Lists of V-Structures grouped by three colored types:
   
   • Black: V-Structures that exist both in the reference structure and the networks from the samples.
   • Blue: V-Structures that do not exist in the reference structure but that appeared in at least once in a sample network. Frequencies are displayed with a negative value.
   • Red: V-Structures that exist in the reference network but that never appeared in any sample network.
4. Comparison structure table: It summarizes all learnt networks from the samples. When identical structures exist, they are gathered.
   
   • First column is the structure identifier
   • The second column is the number of identical structures learnt.
   • The third column represent the frequancy of the whole strucrure : it is the number of times the current structure appreared divided by the total number of structures.
   • The last column indicates whether the reference structure is included or not in the current structure.

• global precision
• relative Gini value
• relative Lift value
• confusion matrices : for occurrences, for fiability and for precision

• Simple network comparison
• Arcs confidence analysis
• Targeted cross validation

1. Navigation bar
   
   • displays synthesis structure
   • displays reference structure
   • displays the previous structure (with regards to current displayed)
   • displays the next structure (with regards to current displayed)
   Navigating through the structures is realized in this order:
   
   1. Synthesis structure
   2. Reference structure
   3. 1st comparison structure
   4. 2nd comparaison structure
   5. ...
2. Display bar
   
   • Zoom in
   • Zoom out
   • Display structure with default zoom
   • Adjust zoom to window size
   • Rotating structure on the left
   •  Rotating structure on the right

• black: Arc or edge that exists in the reference and in at least one comparison network
• blue: Arc or edge that does not exist in the reference network but that exists in at least one comparison network
• red: Arc or edge that exists in the reference network but that does not appear in any comparison network

Settings

Security

• lock the network using the existing password
• remove completely the lock
• change the lock password.

Internationalizing