Glimpse into the world of Artificial Intelligence and Machine Learning
on 16th August, 2017
Have you ever thought, that smartphones that are very dear to us right now, might become obsolete in the next 10 years and get replaced by VR (Virtual Reality)?
Or that if machines could become conscious and start thinking like humans, we might have personal robots rather than personal computers?
Yes, all this could be possible in the near future. Thanks to the Machine Learning and AI revolution, we are quickly moving to techno utopia!
What is Machine Learning?
It is a sub-field of Artificial Intelligence (AI), which simply means improving performance from experience. A good example would be, when a toddler learns to walk, he repeatedly tries to stand and walk with support, sometimes falls down, notices these experiences and improves performance after every step.
Similarly, a machine (computer) learns from data, takes assistance from various algorithms, improves its performance with iterative process and finds out different patterns, hidden correlations in data, which can be later used to perform analysis on new data.
Machine Learning has come a long way. It dates back to the 1950s, when Alan Turing first created the “Turing Test” to determine if a computer has real intelligence. Based on the similar idea, many systems have been developed till date.
Machine Learning became very popular again due to the intersection of Computer Science and Statistics. This synergy resulted in a new way of thinking, the probabilistic approach. In this approach, an uncertainty in the parameters is incorporated in the models. The field shifted to a more data-driven approach as compared to the more knowledge-driven expert systems developed earlier. Many of the current success stories of Machine Learning are the result of the ideas developed at that time.
Big Data explosion and advancement in algorithms have solved the challenges of Machine Learning to a great extent.
How does it work?
The diagram shows the process of creating a model using Machine Learning.
Steps performed while building a Machine Learning Model
- Understand data
- Preprocessing of data
- Feature selection
- Explore different algorithms
- Create a model
- Validate model
- Use model for predictions
Machine Learning Algorithms
Machine Learning is broadly classified into the following two types:
One of the most commonly used machine learning types, in this type of learning, we have input variables (Xs) and output variable(Y). The algorithm used generates a function that predicts the output(Y) based on the input. This is called the supervised method as the algorithm learns from the given X & Y data and when a new data set is given with X1 variables, it will predict output Y1.
In simple words, we have to teach the computer to identify different fruits.
So the data we are going to provide is color, shape, image and name of the fruit. In this case, the characteristics of fruits i.e., color, shape and image become X variables and name of the fruit becomes Y variable. It is called a supervised algorithm because we can supervise the learning process as we know the correct answers, as information about Y variable is present. The algorithm iteratively makes predictions on training data and is corrected. When the acceptable level of performance is achieved, the learning of algorithm stops. When new data is given to the model, it can predict the name of the fruit with good accuracy.
Supervised algorithms include:
- Linear Regression
- Decision Trees
- Random Forest
- Neural Network
In unsupervised learning, we have input variables (Xs); however no corresponding output variable(Y) is present. As the name suggests, the algorithm has to learn from the data and trains itself without any reference or supervision. Algorithms are left to their own devices to discover and present the structure in the data.
In the same example as explained earlier, there would be no labeling, ie, the name of the fruit will not be provided. Hence the algorithm will classify according to the different characteristics present in the data set and classify it accordingly.
Unsupervised algorithms include:
- K-means Clustering
- Hierarchical Clustering
- Gaussian Mixture Models, etc.
Applications of Machine Learning
Machine Learning is all over the place and present across different industries too. It has invaded its way into our daily lives. Below are some of the well-known applications:
- Recommendation Systems – Amazon, Netflix, Google Search Engine
- Business Applications – Customer Segmentation, Customer Retention, Target Marketing
- Speech Recognition – Alexa, Google Assistant, Cortana
- Face Recognition
- Deep Learning
Improve vendor performance by vendor optimization application
|What we did
Based on different performance measures, we identified vendors who are going to become a bottleneck for business in future
Manufacturing companies need Vendor Optimization process to identify vendors with high propensity of failure in adhering to the timelines decided which can affect costs, increasing risks in customer relationship.
The challenge was to identify such vendors as early as possible based on predicted performance measures given by predictive models for some of the key performance indicators like demand and supply, procurement cycle time, delivery performance and return rate , considering benchmarks like OTIF, shipment terms and conditions, safety stock, payment processes and service level.
To account for these challenges, a structured approach was followed to develop a solution:
KloudData Analytics Solution
In the first phase, predictive models were developed using Random Forest, a machine learning technique with maximum accuracy of about 70 percent. Using these predictive models, we are going to find out potential vendors on the basis of demand-supply gap analysis, delivery performance, lead time and quality.
We are also providing a recommendation system to manufacturers to improve the performance of non-performing vendors by providing vendor scorecard on top of their recent performance considering industry standard benchmarks.
Machine Learning has already proven to be an extremely valuable breakthrough in various fields. Thanks to the open source community, Machine Learning is becoming more accessible. While one needs to be wary of this pace of change in technology; it can surely improve our lives in a positive manner.