Originally presented by Andrew Ruegger at Insight Exchange Network's Marketing Metrics & Analytics Summit, this presentation discusses how marketers can leverage machine learning.
They all think they are right, that their method is the way to learn everything with enough data and computational power
Symbolists - Instead of starting with the premise and looking for the conclusions, inverse deduction starts with some premises and conclusions, and essentially works backward to fill in the gaps. The system has to ask itself “what is the knowledge that is missing?” and acquire that knowledge through analysis of existing data sets.
mimicking the human brain, building artificial neurons on AP related to s curves. Google are applying it to areas like vision and image processing, machine translation and experimental neural networks like Google’s Cat Network that helps the computer to recognize cat images.
1964
<<<<ALGORITHM >>>> Genetic Programming - makes an evolves computer programs in the same way nature evolves biologically
Survival of the Fittest & Variation
Evolutionaries - simulating the evolutionary process evolutionaries are applying the idea of genomes and DNA in the evolutionary process to data structures. The survival and offspring of units in an evolutionary model are the performance data.
1964
<<<<ALGORITHM >>>> Genetic Programming - makes an evolves computer programs in the same way nature evolves biologically
Survival of the Fittest & Variation
Evolutionaries - simulating the evolutionary process evolutionaries are applying the idea of genomes and DNA in the evolutionary process to data structures. The survival and offspring of units in an evolutionary model are the performance data.
Bayesians – hot stove - apply a type of “a priori” thinking, believing that there will be some outcomes that are more likely. They then update a hypothesis as they see more data. After several iterations, hypotheses become more likely than others.
-<<<<ALGORITHM >>>> Bayes THEORUM and Derivates <<<<< incorporate new information into out beliefs.
Bayesians – hot stove - apply a type of “a priori” thinking, believing that there will be some outcomes that are more likely. They then update a hypothesis as they see more data. After several iterations, hypotheses become more likely than others.
-<<<<ALGORITHM >>>> Bayes THEORUM and Derivates <<<<< incorporate new information into out beliefs.
Analogizers - making contrast between old and new sets of information. one of the leading proponents of this method, Douglas Hofstadter, in saying that “all intelligence is nothing but analogy.”The master algorithm here, he says, is the “nearest neighbor” principle. Nearest neighbor outcomes can give results that are similar to neural network models. The example of two country models with defined city locations, but with undefined borders. Through the application of the analogy principles, the computer generates a likely border. this “generalizing from similarity” and suggests that it has economic ramifications for technology. One example, he says, is the movie advice technologies that supply movie ratings based on known data sets, where users get recommendations based off of what others have watched previously.
<<<<ALGORITHM >>>> Support vector Machine - figures out which experiences to remember, and how to combine them to make new predictions.
Analogizers - making contrast between old and new sets of information. one of the leading proponents of this method, Douglas Hofstadter, in saying that “all intelligence is nothing but analogy.”The master algorithm here, he says, is the “nearest neighbor” principle. Nearest neighbor outcomes can give results that are similar to neural network models. The example of two country models with defined city locations, but with undefined borders. Through the application of the analogy principles, the computer generates a likely border. this “generalizing from similarity” and suggests that it has economic ramifications for technology. One example, he says, is the movie advice technologies that supply movie ratings based on known data sets, where users get recommendations based off of what others have watched previously.
<<<<ALGORITHM >>>> Support vector Machine - figures out which experiences to remember, and how to combine them to make new predictions.
One day, a female teenager’s father received a Target mailers addressed to his daughter offering her discounts on variety of baby supplies
The father went to target and complained to the local store and their corporate organization.
After doing so the daughter admitted to her father that she was in fact pregnant and had not figured out how to tell him.
Target brilliantly leverages the connected data from their Target credit card and the purchase behavior and online activity associated with that person.
The modeling started to notice highly correlated sequential product purchases.
When the beginning behavior pattern was witnessed by their system, baby mailers are automatically shipped to the name and address associated with the card, and behavior.
They also became aware of the near certainty that individuals who buy Diapers, also buy Beer.
Highest cost -> Overhead cost of employees for Amazon Fresh –Amazons Food Delivery Service
Perishable goods require inspection to ensure expired items are not delivered to customers.
Inspection requires Human Inspectors..
Is there a quantitative to measure a the degradation of produce?
Amazon created a data set of 10,000 strawberries to train machines to identify how many days a strawberry had until it would expire (between 1 – 10 days)
Collected weight, volume, density, 360 imaging to gauge color, grow location, arrival date, and a handful of other metrics.
Had their Quality Assurance inspectors, classify the same strawberry’s on shelf life (1 day to expiration to 10 days to expiration) = The classification Field
Then used Random Forrest (Symbolist) machine learning algorithm to predict how long the life of strawberry would be.
After 2 months of running 1M new strawberries through the model and adjusting some weights, the machines could predict the expiration date with 99.9% accuracy. 1.7% higher than their Quality Assurance Experts.
Amazon used Free opensource software to accomplish this.
Highest cost -> Overhead cost of employees for Amazon Fresh –Amazons Food Delivery Service
Perishable goods require inspection to ensure expired items are not delivered to customers.
Inspection requires Human Inspectors..
Is there a quantitative to measure a the degradation of produce?
Amazon created a data set of 10,000 strawberries to train machines to identify how many days a strawberry had until it would expire (between 1 – 10 days)
Collected weight, volume, density, 360 imaging to gauge color, grow location, arrival date, and a handful of other metrics.
Had their Quality Assurance inspectors, classify the same strawberry’s on shelf life (1 day to expiration to 10 days to expiration) = The classification Field
Then used Random Forrest (Symbolist) machine learning algorithm to predict how long the life of strawberry would be.
After 2 months of running 1M new strawberries through the model and adjusting some weights, the machines could predict the expiration date with 99.9% accuracy. 1.7% higher than their Quality Assurance Experts.
Amazon used Free opensource software to accomplish this.
Highest cost -> Overhead cost of employees for Amazon Fresh –Amazons Food Delivery Service
Perishable goods require inspection to ensure expired items are not delivered to customers.
Inspection requires Human Inspectors..
Is there a quantitative to measure a the degradation of produce?
Amazon created a data set of 10,000 strawberries to train machines to identify how many days a strawberry had until it would expire (between 1 – 10 days)
Collected weight, volume, density, 360 imaging to gauge color, grow location, arrival date, and a handful of other metrics.
Had their Quality Assurance inspectors, classify the same strawberry’s on shelf life (1 day to expiration to 10 days to expiration) = The classification Field
Then used Random Forrest (Symbolist) machine learning algorithm to predict how long the life of strawberry would be.
After 2 months of running 1M new strawberries through the model and adjusting some weights, the machines could predict the expiration date with 99.9% accuracy. 1.7% higher than their Quality Assurance Experts.
Amazon used Free opensource software to accomplish this.
Highest cost -> Overhead cost of employees for Amazon Fresh –Amazons Food Delivery Service
Perishable goods require inspection to ensure expired items are not delivered to customers.
Inspection requires Human Inspectors..
Is there a quantitative to measure a the degradation of produce?
Amazon created a data set of 10,000 strawberries to train machines to identify how many days a strawberry had until it would expire (between 1 – 10 days)
Collected weight, volume, density, 360 imaging to gauge color, grow location, arrival date, and a handful of other metrics.
Had their Quality Assurance inspectors, classify the same strawberry’s on shelf life (1 day to expiration to 10 days to expiration) = The classification Field
Then used Random Forrest (Symbolist) machine learning algorithm to predict how long the life of strawberry would be.
After 2 months of running 1M new strawberries through the model and adjusting some weights, the machines could predict the expiration date with 99.9% accuracy. 1.7% higher than their Quality Assurance Experts.
Amazon used Free opensource software to accomplish this.