## Search found 1241 matches

- Wed May 13, 2020 1:27 pm
- Forum: SMILE
- Topic: Update error
- Replies:
**19** - Views:
**4963**

### Re: Update error

I used this method and the error still persisted. Are you sure you got the same error when inference algorithm was set to EPIS? BTW, Network.SampleCount defaults to 10000 and can be made much larger. Increasing the sample count does not require more memory, but it makes the inference run longer and...

- Wed May 13, 2020 1:24 pm
- Forum: GeNIe
- Topic: Bug(?): DBN ignoring virtual evidence
- Replies:
**1** - Views:
**382**

### Re: Bug(?): DBN ignoring virtual evidence

GeNIe 2.5 has a fix for a bug which caused the temporal virtual evidence to be left unchanged after 'Clear All Evidence' command. This should not directly affect your network, but if you can post it here (or send through a private message) we'll look into the issue. You can also check if the virtual...

- Thu Apr 30, 2020 10:53 am
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

Can you upload your network to Google Drive, Dropbox or similar service and send the link?

- Tue Apr 28, 2020 2:48 pm
- Forum: SMILE
- Topic: Update error
- Replies:
**19** - Views:
**4963**

### Re: Update error

If your evolutionary algorithm produces very complex networks, switch inference algorithm to EPIS (inexact, sampling-based but quire reliable). You may want to adjust the number of samples, use Network.SampleCount property for this.

- Tue Apr 28, 2020 2:46 pm
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

The chain-rule based P(e) is a fallback. Our primary algorithm uses jointree and has running time approximately equal to one inference call, regardless of the number of evidence nodes. The fallback approach is useful when exact inference can't be performed; multiple sampling inference calls will mos...

- Mon Apr 27, 2020 2:46 pm
- Forum: SMILE
- Topic: Update error
- Replies:
**19** - Views:
**4963**

### Re: Update error

Your network is densely connected. Consider reducing the link probability and prior link probability - you have used 0.5 for both. The defaults in GeNIe are 0.1 for link probability and 0.001 for prior link probability.

- Mon Apr 27, 2020 2:36 pm
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

I'm attaching a simple DBN as an example to use with P(e). The xdsl file has one case (use View|Case Manager to apply the case) with three temporal evidence items: Rain(t=2)=true Rain(t=5)=false Rain(t=7)=true P(e) is calculated on an unrolled network (use GeNIe's Unroll command after setting the ev...

- Mon Apr 27, 2020 10:48 am
- Forum: GeNIe
- Topic: interpreting Tornado diagram for sensitive analysis
- Replies:
**3** - Views:
**615**

### Re: interpreting Tornado diagram for sensitive analysis

Yes, that's correct. Sensitivity analysis in GeNIe looks at all parameters, including the parameters in the CPT of target node.So, it is a mutual sensitivity in a sense. Is this interpretation correct?

- Mon Apr 27, 2020 10:46 am
- Forum: GeNIe
- Topic: compare models
- Replies:
**1** - Views:
**526**

### Re: compare models

If you have a dataset which can be used for prediction, use validation and check for accuracy or other parts of the confusion matrix. Use Learning|Validate command to access this functionality.

- Fri Apr 24, 2020 5:59 pm
- Forum: GeNIe
- Topic: GeNIe 2.5 released, includes QGeNIe now
- Replies:
**0** - Views:
**1433**

### GeNIe 2.5 released, includes QGeNIe now

We have released GeNIe 2.5. This version includes QGeNIe, a qualitative interface for Bayesian networks. We have also streamlined diagnostic interface and added new functionality, including: - highlighting in graphical view - "Show Connections" command - "Highlight Paths command - "Select Nodes" com...

- Tue Apr 21, 2020 8:49 pm
- Forum: GeNIe
- Topic: Node inconsistent error.
- Replies:
**2** - Views:
**589**

### Re: Node inconsistent error.

This error is logged during file load when the definition of the node (most likely CPT) is inconsistent; the columns of the table do not sum to 1.0. If you used set_node_definition in your PySMILE program, check the ordering of elements in the definition array. See the "Multidimensional arrays" chap...

- Mon Apr 20, 2020 10:22 pm
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

I believe this may be the issue with some evidence set which contains evidence values with very low P(e). We can check it under the debugger if you can share the network/dataset.

- Mon Apr 20, 2020 10:16 pm
- Forum: GeNIe
- Topic: interpreting Tornado diagram for sensitive analysis
- Replies:
**3** - Views:
**615**

### Re: interpreting Tornado diagram for sensitive analysis

tornado.png Your tornado chart shows how the probability of the PTSD_Alter=High would change if you would start modifying CPT elements in the entire network (including the CPT for the PTSD_Alter node), one at a time. For bar 2, the CPT element is indeed an entry in the definition of the PTSD_Alter....

- Mon Apr 20, 2020 11:34 am
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

Yes, but how do you calculate P(e). Our P(e) implementation should give you the number consistent with chain rule: https://en.wikipedia.org/wiki/Chain_rule_(probability)#More_than_two_random_variables As I wrote earlier in this thread, SMILE first attempts to use a jointree-based algorithm for P(e)...

- Thu Apr 16, 2020 8:56 pm
- Forum: GeNIe
- Topic: Crash/error upon inference in a dynamic Bayesian network
- Replies:
**28** - Views:
**3889**

### Re: Crash/error upon inference in a dynamic Bayesian network

SMILE's EM implementation calculates log likelihood by adding all log(P(e)) for each data row. The value of log likelihood displayed after EM is complete is the number obtained in the last iteration before convergence.