Yes, This Simple WordPress Website in Fontend Has Belief Network Based Power! Rackspace Cloud Intelligence is Not So Easy Graph Than it Appears. We only set the devices ( read as each Cloud Server Instance ) to be configured for capturing data for Cloud Monitoring. It works like a sensor. Actually, if we can o further high-up, each devices can intelligent agents. In that case, we do not have to manually do SSH and kill app or fix the issue. Google is employing machine learning in Data Center, do you think Rackspace Cloud has not done any work?
Belief Network, Machine Learning and Rackspace Cloud Intelligence
Belief Network has lot of names – Bayesian network, Bayes network, Bayes(ian) model or probabilistic directed acyclic graphical model. Belief Network is used for modeling knowledge in biology and computational bioinformatics, medicine, biomonitoring, semantic search, image processing, data fusion, decision based support systems and for lot of stuffs. There are three main tasks for Belief Network :
- Inferring unobserved variables
- Parameter learning
- Structure learning
Suppose, the thing that anesthetists uses – Boyle’s machine (it is a kind of universal name of anesthesia machine, regardless of features, attached equipments, version etc.). How the surgeon will put the patient under anesthesia? Another human is required. But actually, if machine learning is added, it can destroy the job of the anesthetists for the common cases. Boyle’s machine is the best model to apply Machine Learning for example. Just a logical set without any specific programming language :
---
If the patient’s heart rate is < 60 {
// throw error
echo "Your patient can die!";
then give a bolus injection of $this of $this-quantity;
( $this-quantity = $this-value = ($get age + $get sex * xy);
)
where allowed time to get the patient’s heart rate > = 86 is 600 seconds;
{ if patient’s heart rate > = 100 is 600 seconds;
then give a bolus injection of $this of $this-quantity;
do a re-check after 300 seconds;
if patient’s heart rate > = 80 for 600 seconds for 5 time check;
echo “Your patient will not die!”;
elseif (
goto line 1;)
}
end the loop;
}
Easy thing! There are many middle aged surgeons love their anesthetists very much. There are lot of reasons.
1 2 3 4 5 6 7 8 | if function anesthetist exist { do check { gender, age, size, var, } } |
Surgeons exist who ask whether things are going great to the anesthetist. Surgery is a skill, no practical score based exam is never taken (if patient dies, score = 0; else if 70 + appearance; < r u mad?). Which are jokingly said are samples of Inferring unobserved variables, Parameter learning (asking the anesthetist for the next step is based on Gaussian distributions) etc.
Basic thing is - collection of data, analysis of the data; applying logic and automation to eliminate errors. More human on a single work, unfortunately the chance of delay increases after a certain number. The bias is our ego. That thing (the whole stuff) has been written by here :
1 | https://developer.rackspace.com/blog/rackspace-cloud-intelligence-insights-in-monitoring/ |
Belief Network, Machine Learning and Rackspace Cloud Intelligence – I can not see any intelligence!
That simply means, your server’s parameters are not manually tweaked. For example, if we optimize my.cnf with Major Hayden’s script for the database server; we might need to increase the tolerance threshold for memory. For a 2GB server, with well optimized condition, with minor load, 85% RAM consumption can trigger alert. But, you as a human know, you actually left no less than 10% space for real emergency. As it is a virtual instance, with lesser process, we can take such risk. Rackspace actually have a relationship with a provider who develops AI, ML related closed source software. Right now, the level of intelligence is echo "Your patient will not die!"; via email, gradually the whole service will possibly control the nodes to some extent.
