Tag: probabilistic graphical models

# Deep Learning for NLP

Deep learning is usually associated with neural networks.

In this article, we show that generative classifiers are also capable of deep learning.

What is deep learning?

Deep learning is a method of machine learning involving the use of multiple processing layers to learn non-linear functions or boundaries.

What are generative classifiers?

Generative classifiers use the Bayes rule to invert probabilities of the features F given a class c into a prediction of the class c given the features F.

The class predicted by the classifier is the one yielding the highest P(c|F).

A commonly used generative classifier is the Naive Bayes classifier.  It has two layers (one for the features F and one for the classes C).

Deep learning using generative classifiers

The first thing you need for deep learning is a hidden layer.  So you add one more layer H between the C and F layers to get a Hierarchical Bayesian classifier (HBC).

Now, you can compute P(c|F) in a HBC in two ways:

The first equation computes P(c|F) using a product of sums (POS).  The second equation computes P(c|F) using a sum of products (SOP).

POS Equation

We discovered something very interesting about these two equations.

It turns out that if you use the first equation, the HBC reduces to a Naive Bayes classifier. Such an HBC can only learn linear (or quadratic) decision boundaries.

Consider the discrete XOR-like function shown in Figure 1.

There is no way to separate the black dots from the white dots using one straight line.

Such a pattern can only be classified 100% correctly by a non-linear classifier.

If you train a multinomial Naive Bayes classifier on the data in Figure 1, you get the decision boundary seen in Figure 2a.

Note that the dotted area represents the class 1 and the clear area represents the class 0.

It can be seen that no matter what the angle of the line is, at least one point of the four will be misclassified.

In this instance, it is the point at {5, 1} that is misclassified as 0 (since the clear area represents the class 0).

You get the same result if you use a POS HBC.

SOP Equation

Our research showed us that something amazing happens if you use the second equation.

With the “sum of products” equation, the HBC becomes capable of deep learning.

SOP + Multinomial Distribution

The decision boundary learnt by a multinomial non-linear HBC (one that computes the posterior using a sum of products of the hidden-node conditional feature probabilities) is shown in Figure 2b.

The boundary consists of two straight lines passing through the origin. They are angled in such a way that they separate the data points into the two required categories.

All four points are classified correctly since the points at {1, 1} and {5, 5} fall in the clear conical region which represents a classification of 0 whereas the other two points fall in the dotted region representing class 1.

Therefore, the multinomial non-linear hierarchical Bayes classifier can learn the non-linear function of Figure 1.

Gaussian Distribution

The decision boundary learnt by a Gaussian nonlinear HBC is shown in Figure 2c.

The boundary consists of two quadratic curves separating the data points into the required categories.

Therefore, the Gaussian non-linear HBC can also learn the non-linear function depicted in Figure 1.

Conclusion

Since SOP HBCs are multilayered (with a layer of hidden nodes), and can learn non-linear decision boundaries, they can therefore be said to be capable of deep learning.

Applications to NLP

It turns out that the multinomial SOP HBC can outperform a number of linear classifiers at certain tasks.  For more information, read our paper.

Visit Aiaioo Labs

# Building Machine Learning Models that can help with Customer Service and Supply Chain Management

The Laptop that Stopped Working

One fine day, a couple of months ago, a laptop that we owned stopped working.  We heard 4 beeps coming from the machine at intervals but nothing appeared on the screen.

Customer Service

The service person quickly looked up the symptoms in his knowledge base and informed us that 4 beeps meant a memory error.

I replaced first the two memory modules one by one, but the machine still wouldn’t start.  I tried two spare memory modules that I had in the cupboard but the computer wouldn’t start.

I had a brand new computer with me that used the same type and speed of memory as the one we were fixing.  I pulled out its memory chips and inserted them into the faulty computer, but still no luck.

At that point, the service person told me that it must be the mother board itself that was not working.

Second Attempt at Triage

So the next day, a mother board and some memory arrived at my office.  A little later a field engineer showed up and replaced the mother board.   The computer still wouldn’t start up.

When the field engineer heard 4 beeps, the engineer said it MUST BE THE MEMORY.

Third Attempt at Triage

A few days later, a new set of memory modules arrived.

The engineer returned and tried inserting the new memory in.  Still no luck.  The computer would not start and you could still hear the 4 beeps.

A third set of brand new memory modules and a new mother board were sent over.

Fourth Attempt at Triage

The engineer tried both motherboards and various combinations of memory modules, but still, all you could hear were 4 beeps and the computer would not start.

During one of his attempts to combine memory and motherboards, the engineer noticed that though the computer did not start, it did not beep either.

So, the engineer guessed that it was the screen that was not working.  But just to be safe, he’d ask them to send another motherboard and another set of memory modules to go with it.

Fifth Attempt at Triage

The screen, the third motherboard and the fourth set of memory modules arrived in our office and an engineer spent the day trying various combinations of screens, motherboards and memory modules.

But the man on the phone said: “Sir, 4 beeps means there is something wrong with your memory.  I will have them replaced.”

I had to take out my new laptop’s memory and pop it into the faulty machine to convince the engineer and support staff that replacing the memory would not fix the problem.

All the parts were now sent over – the memory, motherboard, processor, drive, and screen.

Sixth Attempt at Triage

Finally, the field engineer found that when he had replaced the processor, the computer was able to boot up with no problems.

Better Root Cause Analysis

The manufacturer could have spared themselves all that expense, time and effort had they used an expert system that relied on a probabilistic model of the symptoms and their causes.

Such a model would be able to tell, given the symptoms, which component was the most likely to have failed.

Such a model would be able to direct a field engineer to the component or components whose replacement would be most likely to fix the problem.

If the attempted fix did not work, the model would simply update its understanding of the problem and recommend a different course of action.

I will illustrate the process using what is known in the machine learning community as a directed probabilistic graphical model.

Run-Through of Root Cause Analysis

Let’s say a failure has occurred and there is only one symptom that can be observed: the laptop won’t start and emits 4 beeps.

The first step is to enter this information into the probabilistic graphical model.  From a list of symptoms, we select the ones that we observe (all observed symptoms are represented as yellow circles in this document).

So the following diagram has only one circle (observed symptom).

Model 1:  The symptom of 4 beeps is modeled in a probabilistic graphical model with a yellow circle as follows:

Now, let’s assume that this symptom can be caused by the failure of memory, the motherboard or the processor.

Model 2:  I can add that information to the predictive model, so that the model now looks like this:

The model captures the belief that the causes of the symptom – processor / memory / motherboard failure are (in the absence of any symptoms) independent of each other.

It also captures the belief that given a symptom like 4 beeps, evidence for one cause will explain away (or decrease the probability of) the other causes.

Once such a model is built, it can tell a field engineer the most probable cause of a symptom, the second most probable cause and so on.

So, the engineer will only have to look at the output of the model’s analysis to know whether he needs to replace one component, or two, and which ones.

When the field engineer goes out and replaces the components, his actions can also be fed into the model.

Model 3:  Below is an extended model into which attempts to fix the problem by replacing the memory can be incorporated.

If a field engineer were to feed into the system the fact that the memory was replaced with a new module and it didn’t fix the problem, the system would be able to immediately figure out that the memory could not be the cause of the problem, and it would suggest the next most probable cause of failure.

Model 4

Finally, in case new memory modules being sent to customers for repairs frequently turned out to be defective, that information could also be added to the model as follows:

Now, if the error rate for new memory modules in the supply chain happens to be high for a particular type of memory, then if memory replacement failed to fix a 4-beep problem, the model would understand that faulty memory could still be the cause of the problem.

Applications to Supply Chain Management

The probabilities of all the nodes adjust themselves all the time and this information can actually be used to detect if the error rates in new memory module deliveries suddenly go up.

Benefits to a Customer Service Process

1.  Formal capture and storage of triage history

2.  Suggestion of cause(s) given the effects (symptoms)

3.  Suggestion of other causes given triage steps performed

What the system will seem to be doing (to the layman):

1.  Recording symptoms

2.  Recommending a course of action

3.  Recording the outcome of the course of action

4.  Recommending next steps

# Frameworks for evaluating values

I recently came across a very interesting 2001 paper by Daphne Koller that dealt with influence diagrams and how they could be applied to game theory.  I came across the paper while doing some background reading on a talk on decision making in accordance with our core values by a friend of mine, Somik Raha.

Influence diagrams are a formalism (very similar to probabilistic graphical models) that are used for making decisions.

What Somik Raha has attempted to do is come up with a framework for making decisions while also taking one’s values into account (either as constraints or as inputs into the decision model).  To do that he proposes extensions to influence diagrams.

What I found interesting when I thought about Daphne Koller’s work and Somik’s together, is that they could possibly give you a framework to evaluate your values.

Koller’s formalism reduces to a game theoretic model, which can be evaluated to determine the outcome of the decisions made by a group of people.

Plug in a formalism based on Somik’s ideas and you just might be able to create a way to quantity the benefits of values.

The Importance of Values?

I have been thinking a bit about values these days because there has been a horrific gang rape in Delhi, and there have recently been numerous incidents of bad driving where friends of mine have been injured in Bangalore.  Then there is corruption.  Our society seems to be quite happy with inequality and vast differences in the distribution of wealth.  It make me wonder if our values are to blame.

I have often wondered whether some of our problems originate in our value systems and whether the value systems that we consider sacrosanct in India are really very good ones.

Let me take just a couple of values that most Indians would consider to be very good values

1. Non-violence
2. Obedience

and let’s discuss them in more detail.

1.  Non-violence

This value appeals not just to people in India.  You see variants of the value of non-violence appear in Tolstoy’s writings and in Semitic religions, as you can see from the Bible (“turn the other cheek”) and the Quran (“give alms to one who begs from you, even if he comes on the back of a horse”).

The issue with this sort of value is that it makes a person (and those around him/her) extremely vulnerable to injustice.

In India, we restrict the liberty of women – in their choice of clothing, company and lifestyle – for fear that they could be in danger if they violated societal norms.  This shows that none of us want to fight society or cross swords with someone who might make disparaging comments about personal choices.

Moreover, possibly as a result of the value of non-violence, very few Indians if any are taught fighting skills in school.  So, even if a person really wanted to act, say to protect a friend, he or she might not really have the skills to take down an aggressor.

So, instead of protecting and standing up for people who might be vulnerable, we become their tormentors and make their lives more miserable, just so we don’t have to get our hands dirty, or because we don’t have the skills and strength to do squat.

I’ve written about how bribes are openly collected by traffic policemen.  It should be very easy to put a stop to such behavior if you’re willing to fight.

If non-violence is not a core value, then how do we protect people from tearing each other to bits?

We could start with a question like:  non-violence for what purpose?  (turning it into an extrinsic value)

If the answer is something like, “so that the weak feel protected”, why not make protecting the weak our core value?

I’d prefer teaching kids values like “Don’t ever turn your back on a bully” rather than values like “Don’t fight anybody, and just come home safe, child!”

2.  Obedience

Indian parents love to boast that their child is “such an obedient child!”

Is that a good thing?

Obedience is different from politeness or respect.  The latter are mutual but the former is one way.

So, the politics of obedience creates a hierarchy of subservience.

In India, Parents expect complete obedience from Children.

The Police expect complete obedience from People.

The Politicians expect complete obedience from Police.

Teachers expect complete obedience from Students.

Managers expect complete obedience from Employees.

The creation of the hierarchy (through expectations of obedience) can be very dangerous in many ways.

1.  It can stifle creativity and problem-solving ability.  There is a bias against ideas flowing up a hierarchy because those higher up the hierarchy claim their place above those required to be obedient to them on the premise that they are somehow superior to those below them.  A good example is how parliament will not accept that people have a right to demand a bill against corruption (members of the Indian parliament claim that parliament is supreme in a parliamentary democracy – not the citizens that the parliamentarians represent).

2.  It can leave young people ill-equipped to defend their personal spaces.  I read in a paper on rape that many rapists approach victims by testing their boundaries.  They make comments and otherwise violate the intended victim’s personal boundaries.  If these are not strongly resisted, the probability of an assault becomes greater.  Another strategy used by rapists is to move their intended victim to a new location where they are more vulnerable. It is very important for people to be conditioned so that they do not obey an order by an attacker to relocate under any circumstances.

3.  The hierarchies perpetuate the power of stronger (bigger, older or richer) parties by providing social sanction to their dominant position, and so hinder social mobility.

4.  The obedience hierarchy could allow a few people at the top to amass too much power. It might have, for example, prevented cops from disobeying those in power during the Gujarat riots.

5.  Obedience means valuing rules above truth.  Obedience implies not challenging the rules or the status quo.  So there is little scope for discovering if the rules really are good ones for everybody.  People often defend something they assert with a “because I said so.” – that is, you are expected to believe them because of their authority, and not because they can substantiate their assertion.

Obedience as an absolute value is not entirely harmless.  It could be dangerous to us as a society because corrupt politicians can use the pliability and obedience of people around them to get away with evil (remember the activist who was hacked to death on the orders of a corporator from Bangalore, the journalist who was burnt to death in Uttar Pradesh, or the shutdown that the former Chief Minister of Karnataka State ordered when he was about to be investigated for corrupt dealings?).

I’d love to replace “obedience” with something else, perhaps “honesty” and “trustworthiness” and “pride”.

Summary

I understand that we as Indians are very proud of our values but I’ve tried to argue that our values need to be re-examined.

Personally, I’d love to see the day when we replace all our values with just the value of trustworthiness.

Trustworthiness as a value would mean we’d fight for each other.  It would mean we’d protect the weak.  It would mean we’d be on time.  It would mean we’d be honest.  It would mean we’d be capable and skilled and strong.  It would mean we’d be proud of each other.  It might mean we’d never lose another war.

Reading Koller’s and Somik’s work you get the feeling that one day you might be able to evaluate the comparative benefits of two sets of values, and pick the better one, using plain math.

And hopefully, by showing them mathematical proofs, you can convince people to change their values and pick better ones for themselves.