Kenny Bastani's Blog

The Heuristics of Turning a Thought Into Work

2011-11-16T23:43:00.001-05:00

A few moments ago I somewhat randomly conceived the notion that consuming food regularly is an important physical action for productivity of the physical body to at the very worst survive and at the very best improve bodily function. While of course I know this when rationalizing the necessities for life, it is due to the random nature for which this has so often traveled to the forefront of my mind that has intrigued me so.

It intrigues me because we take the mind for granted.

This is largely due to the fact that, as conscious intelligent lifeforms, it seems that we have control of our mind. We may have conscious control of our physical actions, however the manner in which relational concepts pop into the mind are largely out of our conscious control, with the manner referring to the content and order of the relational concept.

The subconscious mind is the content delivery network of the human brain. Consider it your brain's own Google.

Last year Google released the instant search feature on their search engine, which eliminated the criteria to press the enter key to submit your search.

Let's go over the heuristics of the primary use case, better known as a scenario, for Google's search feature. As you are typing the letters of your search query into the search box there are two primary actions being triggered in parallel:

Google automatically completes your search query based off of highly relevant search patterns.
After each letter is typed, relevant media content, à la search results, are instantly rendered on the search page.

Why is this relevant to the topic?

Because you take Google's magic for granted.

Why aren't you thanking the Google gods every time you perform a search? Why aren't you mentioning thanks for Google's search feature before eating supper, for which the knowledge of its recipe, ingredients, prices, and nearest store locations you would have otherwise been without.

Think of your life without it! You would be completely disabled in this modern era from performing regular life activities related to traveling, shopping, working, leisure and so much more.

If Google's search is SO important to the fitness of the modern Homo sapiens, why is it that we regularly take it for granted?

It is because Google's search is seemingly effortless. It never communicates to you its effort or the effort of the engineers who designed it, built it, and continually improve it.


Rosie never complains about doing work

I present to you the humble servant. It is always those who withhold mention of their effort level that are most often taken for granted. In other words, those who don't complain about performing work perform their work effortlessly.

The Genesis creation narrative provides an excellent example. It states that God created the heavens and the earth in 6 days. There is no mention of God's work effort for those first 6 days, it does mention that on the 7th day God rested from His work, having completed the heavens and the earth. The mention of resting on the 7th day implies that God's work creating the heavens and the earth was not effortless.

If we could ask God about how much effort it took to build the heavens and the earth, God would probably tell us something like "It's all good, man."

If you need another narrative, "The Ungrateful Son" by the Grimm Brothers illustrates the concept nicely.

Do you or others you know take things for granted? Has a person, whether they be an employee, significant other, friend, or even a politician at some point in your life told you that you are taking them or something else for granted? I'd wager to bet the answer is a resounding yes.

This is all good and nice, but what does it have to do with the subconscious mind? The effort of the subconscious mind is seemingly effortless and therefore we take it for granted. It indirectly triggers your brain's enormous network of neurons to fire in areas that are most relevant and important to your actions. Which is why when I started to feel that empty feeling in my belly, I was suddenly and magically reminded that earlier today I brought home extra food from Starbucks's bistro selection and it is ready for consumption without any effort to prepare it.

This information popped into my mind within a matter of seconds after my belly grumbled. That is some useful information. I didn't have to spend any effort probing my mind for that information. It is as if I had Googled the word "Hungry" in my brain and the latest news results contained a video of me putting the food in the refrigerator earlier in the day and an associated news article that is a persuasive piece on how delicious and effortless it would be to retrieve that food and consume it.

The following is the result of a thought experiment on how a Google search might be adapted to work in an AI system that employs the use of neural networks.

Neural network markers for subconscious emphasis on primary directives

The subconscious mind will encode reference markers within neural network hubs that are connected to the primary directive modules of a physical activity. The reference markers are continually moved to the most relevant areas for quick content analysis preceding the desire to act.

When the conscious mind triggers a primary directive with a desire to act, the reference markers placed by the subconscious mind will activate within the network hubs where they were earlier placed.

The activation function will cause the conscious interpreter to suddenly bring the relevant hub's data forward in a seemingly random order to an interpreter module, which will perform a function to retrieve associated context that describes attributes of the received network data.

Once thought activity has stabilized below an activity threshold for the primary directive module, the subconscious mind will further activate reference markers for physically performing the activity response that was chosen by the conscious mind based on the hub's relevant context data that was earlier received by the interpreter module. This will cause another set of reference marker activations for conceptualizing and performing consequential physical action in response to environmental changes during activity.

For more neuroscience behind this thought experiment, read this research paper that was recently published in The Journal of Neuroscience: Rich-Club Organization of the Human Connectome

Linear and Expanding Irrational Universe

2011-11-02T03:55:00.000-04:00

For many years now I have been working on a grand unified theory of physics that explains the universe in all its glory. After much consolidation, I've summarized the full theory down to 14 points that detail it in words, without any mathematical formula. I should mention that this is a radical theory that rewrites physics from the ground up rather than from the sky down. Its purpose is to provoke thought, and as such it is not in any way attempting to undermine the hard working and brilliant community of scientists who dedicate their lives to understanding the universe.

Pythagoras and Rationality

Pythagoras, whose sculpture is featured to the left, used rationality to explain the universe.

"In this theater of man's life, it is reserved only for God and angels to be lookers-on." -Pythagoras

In this quote, Pythagoras is stating that there are two perspectives of the universe; the internal perspective of man and the external perspective of a creator. God's perspective is above man, as man's perspective is below God, which to Pythagoras forms the basis for rationality; a divine ratio.

Theory of a Linear Irrational Universe

A few words to any brilliant theoretical physicists, as well as any notorious practitioners of cynicism towards crackpot grand unified theories: Please be gentle.

1. The universe through all of time is an expanding linear sequence of irrational numbers.

2. All of the substance that is contained within the universe is interpreted by the human brain as distinct recognizable groupings of patterns on a linear sequence of irrational numbers.

3. The third dimension that is perceived by the human brain is a visualization of the linear sequence of irrational numbers, which creates an illusion of higher dimensions through a form of organic multiplexing.

4. The third dimension perception is an evolutionary trait that gave species a genetic advantage, as evolved sensory became a necessary element of survival for multicellular organisms.

5. The flow of information moves linearly in one direction, with the speed of light being relative to the length of the entire sequence.

6. Quantum entanglement is the observable effect of a group of upstream irrational numbers affecting a specific group of downstream irrational numbers in a predictable and targeted manner.

7. Each irrational number in the sequence sheds a portion of its value downstream during a full iteration, with the last irrational number in the sequence serving as the capstone number, which sheds a portion of its value to create a new irrational number at the start of the sequence.

8. A black hole is formed by the increasing build up of high density irrational number groups on the linear sequence, for which gravity is caused by the forward pressure generated while irrational number groups are traveling downstream towards the denser cluster of irrational numbers.

9. All empty space in the universe is made up of relatively small values of irrational numbers that exist between relatively large values of irrational number densities.

10. When energy, which is the process of irrational number groups shedding value downstream in a wavelike pattern, approaches relatively high value irrational number groups that are downstream, the pressure buildup causes fusion of irrational number groups into new and distinctly denser patterns of irrational number groups, which forms elements of differing mass.

11. Time is perceived as the rate of growth and expansion of the linear sequence of irrational numbers increasing in length during each cycle. As the length between the first and last irrational numbers in the sequence grows, so to does the time it takes to compute one full linear and sequential cycle of downstream irrational number flow, for which the last tick of the cycle expands the sequence by one new irrational number at the start of the linear sequence.

12. Light waves in the third dimension are perceived when irrational number groups come into contact with the eye's retina and are then organically multiplexed for interpretation by the sensory system of an organism. Color and luminosity are interpreted as downstream energy flowing from an upstream radiating density of irrational number groups, creating waves of differing lengths that travel at or near the speed of light.

13. During each tick of a full cycle of downstream energy flow, a portion of each irrational number's value is sequentially shed to the next downstream irrational number in the amount of the upstream irrational number's full value divided by Pi.

14. At the final tick of a full cycle the last irrational number in the sequence, referred to as the capstone number, sheds a portion of its value in the amount of its full value divided by Pi, for which the resulting value that is shed is used to create a new irrational number of that value at the start of the linear sequence, thus with each cycle the size and length of the universe expands while always maintaining a constant total energy from the first cycle, whereas the first cycle would be the original big bang event.

Capstone Number?

The capstone number is the last number of the sequence, in which its primary function is to create a new irrational number at the start of the sequence at the last tick of each cycle. I've named it the capstone number because both its function and position matches closely the definition of an actual capstone.

cap·stone

1. The top stone of a structure or wall.

2. The crowning achievement or final stroke; the culmination or acme.

Video Simulations

The video below demonstrates a time sequence of the expanding set of irrational numbers over the course of 100 cycles of activity. The sequence expands to 100 irrational numbers from 1, but always maintains a constant total sum of 1. As the universe expands outward, energy is shed backwards, for which the amount of energy is represented on the Y-Axis.

This next video is a visualization of my irrational universe's big bang event that improves the video output of the digital radioactivity sensor. This would be a the surface texture of a 360 degree panorama.

The universe's expansion is irrational, however, I do believe that intelligent life may indeed perceive the universe in a rational and harmonic way. Why else would we have the incessant need to organize it?

A Skilled Laborer's Economy

2011-10-08T19:10:00.000-04:00

For our nation's economy to improve, the nation's productivity must improve. For productivity to improve, demand must rise for products. When demand is low, new products that yield a higher demand must be designed and produced for market consumption. New products usually demand new skills, and as the complexity of a product grows, so too does the skill set required to work with it or on it.

Skill sets required to innovate in competitive markets require education and a vast amount of work experience. When job growth decreases, available skill sets for jobs requiring work experience also decreases. Employers in innovative competitive consumer product industries rarely want to roll the dice on educated laborers who lack work experience in their field. They need people who know the market, the product, and the trade.

The following bullet points are the results of the thought experiment, in which I used causality to rationalize each event as a logical outcome and cause.

The economy's shape is unfortunately a product of increasing technological complexity, requiring skilled laborers to fill roles.
The complexity is so high that work experience using these skills becomes more valuable than education.
Skilled laborers end up settling for entry level positions, and highly educated laborers are forced to take positions in the service industry.
This has a domino effect in which neither skilled nor educated laborers are able to find roles they used to be able to fill.
Now some students who don't settle for the service industry after graduating, they end up deciding to wait it out and go back to school.
In this process, they think that even more education increases their job opportunities. They end up borrowing more money.
These students then graduate largely in debt, still with no work experience while other laborers became more skilled.
Now these students who have 2 or 3 degrees, including a masters, end up being forced to take those service industry jobs to pay the loans.
When the media starts to realize this, the value of higher education will implode. Banks will stop lending. Financial aid will evaporate.
Instead of high school grads going to college, they stay home and compete with highly educated laborers for jobs, burdening the middle class.
With more mouths to feed, the young jobless laborers become frustrated and angry, they turn to blame the government or corporations.
That is why the protests happening at this time in the U.S. are significant. As the national job pandemic grows, so to will the number of frustrated protesters.

Object Recognition in Still Images

2011-05-30T00:33:00.000-04:00

Object recognition is the foundation of how we judge the intelligence of an entity.
Let's start with a few assumptions on the human visual cortex:

The human mind receives sensory streams of a normalized format.
The human mind usually does not encode an entire visual scene into memory.
The human mind only encodes abstracted criteria based on frequency of recurrence.
Criteria is hierarchical with the first criteria being light (as oppose to dark).

In the development of the human brain to perceive and separate objects in an environment, the sensory organs are a non-stop stream of raw data.

Imagine yourself sitting atop a rock near a stream of water. You are interested in catching a few fish as they travel down stream. You harness a net and fashion it to catch any object the size of your desired fish or larger. You carefully place the net into the stream so that it filters the maximum amount of water, as not to miss any fish. Objects start to get caught in your net, as expected.

Within thirty minutes, you've caught 4 rocks, 5 toads, 3 crawfish, and 4 fish. Because the criteria of the filter only specified a maximum size, you caught more than just fish. For the net to understand what a fish is, in comparison to the other objects in the stream, you need to find similarities between rocks, toads, and crawfish and design the net to let them pass through unharmed.

This is how the mind works. The misconception is that we believe we see the world for what it is. That's not true and I dare you to prove me otherwise. Your sensory organs provide streams of data to your mind and you apply criteria in a stack until a familiar pattern emerges. The data falls within a range of perception. Dark to light. Quiet to loud. Tasteless to flavorful. So on and so forth.

Herein lies the problem that has inhibited the human mind to create true artificial intelligence. We are still building the nets to catch the fish. Machines need to be able to create and alter their own fishing nets.

In example, I've created an algorithm for visual object recognition. I've provided it with the following image of shapes as a sensory stream:

I've designed the algorithm to determine what criteria splits the image into two categories. The objects of interest all belong to one of those two categories. Since each pixel has a color, we can determine the brightness of each one. The key to object recognition is eliminating background noise to observe relevant patterns. By eliminating all criteria that surrounds a familiar object but does not include it, the program learns how to create complex filters to identify objects.

As the algorithm has no working memory at this time, it cannot go much further in creating nets to identify what the objects in the image are. The next step is to filter the image into a format that will allow the algorithm to identify similar objects from other image streams. To do this, I've created an algorithm that "shakes" the image stream to find the edges of the filtered objects. You can see the result below:

As the algorithm learns to improve its nets, the criteria will become more defined and help identify the objects in complex scenes.

I ran the same algorithm on an actual photo of people:

Raw stream:

After one iteration of splitting by criteria:

Neural Networks for Identifying Behavior

2011-04-04T13:07:00.000-04:00

There is a great history on machine intelligence. The brightest individuals on the planet have been on an unending search for a means to replicate our human intelligence, to perform tasks on par or better than the average human. Imagine the possibilities.

This is the holy grail of innovation. If we can teach machines how to be intelligent they can efficiently do trivial and boring repetitive work without complaint or error. Yet, despite almost boundless advancements in the field of computer technology, R2D2 is still just a science fiction fantasy.

The simple answer to why is that we've built machines to interpret instructions. Even if you teach a machine to give instructions, they are still interpreting your instructions to give those instructions.

So is it even possible to make an application that is intelligent on a digital infrastructure that simulates the human brain?

Yes, I believe so.

While a computer might be constrained to the world of mathematics instead of physics and biology, it is very good at simulating the two latters. With advancements in functional programming, we are moving in the right direction.

How would one build a smart tweet filter that filters on behavior?

To do this one would need to create a simulated processor that is analogous to the human brain.

So how does the brain interpret and give instructions?

From what science has so far uncovered, electrical impulses through dense networks of neurons. So in essence, we need to build the digital counterpart to a biological neuron.

Luckily for us, the groundwork for neural networks has already been laid out. So how do we use a neural network to develop a means to filter tweets on behavior? Like the human brain, it is easy to understand what a neural network does but not how it works.

We don't speak brain, at least not yet.

How a neural network works

A neural network is a composition of digital neurons that take inputs and provides outputs. Each neuron has an adjusted weight that needs to be exceeded by the input to cause the neuron to activate. The activation of the neuron will cause the input to be sent to the next layer of neurons connected to the activated neuron.

The weights associated with the neuron allows one to train the network to achieve a desired result.

So how can we train those neurons to filter on behavior?

By teaching it like you would teach a human. You provide it with examples and tell it what behavior is desirable and what behavior is not.

There is one simple requirement to be able to do this, you and the network have to be speaking the same language. Teaching it how to speak English could take some serious time, so let's invent a language that will take less time to teach.

Communication based training

To train the network, it is mandatory that we speak a language that the network understands. To communicate with the network, we have to have words that demonstrate a few things:

Desirable
Undesirable

How desirable
How undesirable

Filter
Do not filter

This is a good start. We have defined a few polarities to communicate to our network what it is we want or do not want. However, we need to go a step further. We have to answer the question "Why?"

Why this is desirable
Why this is undesirable

Why filter this
Why do not filter this

If we expect our machine friends to understand the instructions we are giving, we can't be mad at them for asking why. In fact, if the network has context as to why, it can better identify the desirable or undesirable behavior.

Once the network is speaking the same language, it can be trained, and it will output a full spectrum of categorization.

Full spectrum categorization

Imagine that the neural network has mapped out behavior linearly from point A to point B and that it has categorized the spectrum into different behaviors. Let's also say that the first half of the spectrum, closest to point A, are desirable behaviors, with the most desirable behaviors closest to point A.

As we move away from point A, behaviors become less desirable. The least desired behaviors on the spectrum are stored closest to point B and become more desirable the farther they are from point B.

While the spectrum is produced from training the weights, it is useful to store in memory examples of behavior so if the spectrum goes through a shake-up, it can reassess and reclassify behaviors. Just like a human.

Conclusion

If we expect a certain level of intelligence from machines, we have to understand that learning is a process, just like a human. Machine intelligence is prone to error just as humans are prone to error, because reasoning is not an absolute science.

Machine learning is a helpful supplement to deterministic algorithms. Just as we post-process the output of humans on Twitter with deterministic algorithms, we can do the same with the outputs of our trained neural networks.

While this article is purely theoretical to this point, it lays the foundation for what a capable neural network could do with full spectrum categorization.

Why Delivering Results is Setting the Bar Too Low

2011-03-28T19:45:00.000-04:00

How do you really achieve success when delivering a solution to a customer?

You don't truly deliver until you've made your customer happy, and I don't mean just happy, I mean ecstatic with joy over the quality of your extra effort and overall work.

Once upon a time I would have said that delivering results was what made me a success. I’ve since learned that delivering on an expectation is a missed opportunity to exceed expectations and generate some smiles from your hard work in addition to dollars.

When you deliver results to the customer, they are simply satisfied. To make them happy, you should listen to them. When you listen to them, and I do mean all of them, not just the executives of an organization but their entire workforce, you are being given high level directions on what it is that makes the customer as a whole happy with your solution. If you pretend to listen, because you think you have everything covered to deliver, you are missing the chance to deliver happiness and create a fruitful relationship that generates business on both ends.

There will always be times that the customer does not understand or agree with you, even though you feel you are right. If your number one goal is to just deliver, then you may feel that complying with the customer’s logic is counter-intuitive to delivering results as quickly as possible.

More often than not, the customer will hold their ground if you attempt to further clarify your streamlined and possibly confusing plan of delivery. In the customer's mind, you are cutting corners to save on costs. This is frustrating, and while the easiest path may seem to somehow escalate and circumnavigate the unnecessary and perhaps pointless demands of an unruly customer or stakeholder, you have to teach yourself that this is just a part of the challenge to exceed expectations.

Lowering production costs by cutting corners was the norm in the consumer mobile devices market before Apple released the iPhone. Quantity over quality ain't gonna cut it anymore, you need to get your hands dirty and deliver some quality solutions.

Enhancing quality of your solutions may add more work but I guarantee that extra work will go much further than the work it takes to “just deliver”. In the long run you'll earn more money from return business than if you are just focused on production.

It helps to remember that everyone you serve is your customer. If you consider all of the people that you work with as your customer, you will always succeed at landing smiles.