• 18 Dec, 17
• by niaja
• in Newsletter
• 0
• 2

The Machine Learning Career Path

By: Vanessa Avila

What is Machine Learning?

For those wondering what machine learning is, its name is quite self-explanatory: a machine that learns. Machine Learning is not a new discovery; It has been known to exist since the 1950s. It stemmed from the grounds of artificial intelligence alongside the interpretation that our brains are essentially computers. In Gary Donovan’s article, “7 Machine Learning Pioneers Timeline,” he writes about Arthur Samuel, one of the early pioneers of machine learning who created the first program where the machine’s purpose was to learn and improve on its own. This was exhibited by programming a machine to play checkers. As the computer progressed throughout the game, it managed to discern good and bad moves, and improved by broadening its search tree (Donovan, 2015). The program allowed the computer to learn from experience, as humans would. This groundbreaking creation defied the ideology that programs are instructions meant to be highly specific, rigid, and inflexible for a machine to perform a particular task successfully. By allowing a computer to learn from progress, it was evident that it was possible to program a machine to a certain extent and have it figure out the rest of its tasks by itself, unsupervised.

 

The Identity of Machine Learning and Tips for Starters

Since machine learning is like a condensed variety of important topics, it’s essential for one to know what to learn and how to get started. The best and well-known fields that could give a student a step ahead in machine learning occupations are particularly Data Science or Computer Science. Machine learning’s entity is a combination of both. Another important area of machine learning is the realm of “deep learning,” which, as seen in the Venn diagram below, is a subset of machine learning that is concerned with algorithms that imitate neural activity (also generally known as neural networks.) Although there exist debates in which Venn Diagram is best to visually represent the differences of the fields, the Data Science Venn diagram by George Piatetsky-Shapiro (President of KDNuggets, the leading website that focuses on Data Science, Data Mining, Machine Learning, Analytics, and Big Data),  is used for reference:

In the statistical aspect of things,  this career path involves plentiful data, and with better technology at present time, huge amounts of data are stored on a daily basis. Specifically, in the realm of “Big Data” (a term used to describe a huge volume of data to be used for analysis (SAS, n.d.)) machine learning is involved with Data Science to analyze these huge collections of information. Data mining particularly uses machine learning to apply to Big Data (IO Technologies Inc., n.d.), and that is how machine learning partially submerges itself to the data science and the data mining world.  The huge data input can then be observed for patterns, which kicks in the engineering application of pattern recognition. This general gist of what goes on in the ropes of machine learning is essential for the making of inferences from a huge sample that leads to the creation of predictions or predictive models.

Hand in hand with math and engineering, computer science is also important in machine learning. Author of the book Introduction to Machine Learning, Ethem Alpaydin (2014) states  “Machine learning uses the theory of statistics in building mathematical models… The role of computer science is twofold: First, in training, we need efficient algorithms to solve the optimization problem, as well as to store and process the massive amount of data we generally have. Second, once a model is learned, its representation and algorithmic solution for inference need to be efficient as well,” (p. 3). Computer science is also connected to the very important branch of artificial intelligence in the machine learning process. since it is needed for artificial neural networks. Artificial neural networks are responsible for processing information similarly like how our brains process information (Kumar et. al, 2010). It allows for adaptation when changes occur, giving it the ability to learn without giving every solution for every possible situation (Alpaydin, 2014, p. 2).       

There is more in-depth information to be shared about machine learning, but there have been a few bits highlighted for anyone who is willing to have a take on the career. In William Terdoslavich’s article “Real Skills You’ll Need for A.I. and Machine Learning,” software engineer at Infer, Joel Dodge, mentions that it is uncommon to find applicants who have years of machine learning experience, so to fix that issue, most are expected to efficiently do a lot of on-the-job learning. Abdul Razack, head of platforms and senior vice president of Infosys, also shares to Terdoslavich an alternative hiring approach of taking specialists of a different field and teaching them the skill set needed for machine learning. Additionally, head of machine learning at Bloomberg, Gary Kazantsev stresses that subjects such as probability, statistics, linear algebra, and mathematical optimization are all important for those wishing to develop their own algorithms (Terdoslavich, 2016). It is essential to remember that a degree and a school certificate is a small fraction of what the hirer weighs in his or her mind when it comes to the hiring process. The ability to showcase one’s potential in the workplace is also important. If anything, it is of bigger weight to be taken into consideration.

In a more experience-based context for machine learning,  Vincent Granville’s blog post “Data Science and Machine Learning Without Mathematics” describes the not-so-math-heavy techniques of his day-to-day tasks. He begins to enumerate several helpful articles he has written which further describes his machine learning job activities on a day-to-day basis that do not require heavy math use: variance, clustering, and density estimation, model-free confidence intervals, and more. He mentions that basic excel skills helped in advanced machine learning and can aid in the implementation of important techniques involved in a machine learning career (Granville, 2017). So for anyone who wants to get ahead, mastering excel is a great place to start!

Many individuals self-learn machine learning and a lot of content online are provided to get an idea of starting. Eric Jang, a research engineer at Google Brain, claims that there’s no actual way to learning this field rather one should come up with study methods that the he or she is comfortable with. He also suggests specializing and choosing one sub-field of Deep Learning (in which he lists down) as a trick to be efficient in a specific aspect of machine learning (Quora, 2017). For people who want a more organized walkthrough for independently studying the career, sites like https://www.machinelearningmastery.com and https://www.elitedatascience.com offer step-by-step procedures, as well as other resources, such as free textbooks available for reading.

Machine Learning Applications

As mentioned previously, machine learning isn’t something people converse about casually.  However, it is far common in everyday use than people realize it to be. It’s not a surprise that a person can encounter typical machine learning in a day. A familiar name that people know is “Siri,” who of which is a great example of exhibiting the applications of machine learning. Virtual personal assistants like Siri need machine learning to gather the information of how the user previously interacted with them, which in return gives the user a more refined accuracy in results based off of the user’s preferences (Daffodil Software, 2017). Using search engines such as Google’s also involves machine learning since the program watches search activity and interprets how the user interacts with the search results. From there the program learns to render better search results that will fit more accordingly to what a user is searching for. Other applications, among many, involve fraud detection, smart cars, financial trading, and even your simple and almost present recommendations from Netflix (Marr, 2016)!

The list can drag on for common machine learning encounters when it comes to business and technology, what with the main statistical and computer science requirements to fit the occupation, but believe it or not, machine learning is also excelling in the world of medicine. In the article “7 Applications of Machine Learning in Pharma and Medicine,” Fagella writes about a study that estimated a generated value of up to $100 billion annually in the pharmaceutical and medical industry with the help of machine learning. It was mentioned that the healthcare sector has been great in accumulating data from research, development, patients, caregivers, physicians, clinics, and more. This paves the way for machine learning to help analyze the collected data to aid the prevention of diseases as well as the treatment of individual patients. Enumerated were particular aspects of this industry that would work efficiently with machine learning. To name a few,  disease diagnosis, drug manufacturing, drug discovery, clinical trial research, radiology, and behavioral modification are all of the areas that perform with better efficiency with machine learning. However, most of these examples are still works in progress and research and experimentation is still an ongoing task to pave way for machine learning to do its magic (Fagella, 2017).

Machine learning can also be applied to social issues. In an MIT’s Technology Review article, students from the University of Rochester managed to create an algorithm that trains the machine to detect alcohol-influenced tweets and to use this set of data to monitor alcohol activity. They filter tweets that mention alcohol or any word that might be a connotation of an alcohol-related activity. This work also has another inspiration and that is to locate the Twitter user to determine whether they are drinking at home or not. A list of filtered words that would signify if a user was in their own vicinity is again used to identify the user’s location. Nabil Hossain and his fellow co-workers who are hard at work in this project believe that this method can significantly affect how we respond to the health problems that are caused not just by alcohol, but other activities that could cause a public health issue (Emerging Technology From The Arxiv, 2016).

It is evident that machine learning is an integral part of today’s society. People who are aware of machine learning existence know that the applications that do exist and the ones that will exist in the future will be very useful alongside the technological advancements the world has yet to provide. A viable reason as to why its existence is a little minimized from common knowledge is because it is a career path that highly interconnects other fields. However, this characteristic shows that a machine learning career is diverse in terms of the subjects involved in being successful in this career path. The options of where a person wants to practice his or her machine learning expertise is plentiful. Individuals who are uncertain of which path to pursue but have spread interests in the fields of STEM might find machine learning to be a career of interest and should definitely consider looking into it!

Sources:

Alpaydin, E. (2014). Introduction to Machine Learning. Cambridge: MIT Press.

Daffodil Software. (2017, July 31). 9 Applications of Machine Learning from Day-to-Day Life.

Retrieved October 27, 2017, from https://becominghuman.ai/9-applications-of-machine-learning-from-day-to-day-life-112a47a429d0

Donovan, G. (2015, May 5). 7 Machine Learning Pioneers Timeline. Retrieved October 27,

2017, from http://techjaw.com/2015/05/05/7-machine-learning-pioneers-timeline/

Emerging Technology from the arXiv. (2016, April 25). Machine-Learning Algorithm Identifies

Tweets Sent Under the Influence of Alcohol. Retrieved October 23, 2017, from https://www.technologyreview.com/s/601051/machine-learning-algorithm-identifies-tweets-sent-under-the-influence-of-alcohol/

Faggella, D. (2017, September 01). 7 Applications of Machine Learning in Pharma and

Medicine. Retrieved October 23, 2017, from

https://www.techemergence.com/machine-learning-in-pharma-medicine/

Granville, V. (2017, June 26). Data Science and Machine Learning Without Mathematics [Blog

Post]. Retrieved October 30, 2017 from https://www.datasciencecentral.com/profiles/blogs/data-science-and-machine-learning-without-mathematics

IO Technologies Incorporation (n.d.). What is The Difference Between Data Science, Data

Analysis, Big Data, Data Analytics, Data Mining and Machine Learning? Retrieved October 27, 2017, from https://onthe.io/learn/en/category/analytic/What-is-the-difference-between-Data-Science%2C-Data-Analysis%2C-Big-Data%2C-Data-Analytics%2C–Data-Mining-and-Machine-Learning%3F

Kumar, J. B., Bhattacharyya D., Kim T. (2010). Use of Artificial Neural Network In Pattern

Recognition. International Journal of Software Engineering and Its Applications, 4(2), pp. 23-34.

Marr, B. (2016, September 30). The Top 10 AI And Machine Learning Use Cases Everyone

Should Know About. Retrieved October 20, 2017, from

https://www.forbes.com/sites/bernardmarr/2016/09/30/what-are-the-top-10-use-cases-for-machine-learning-and-ai/#11d1b87394c9

Mayo, M. (2016). Data Science Venn Diagram [Digital image]. Retrieved October 29, 2017,

from https://www.kdnuggets.com/2016/03/data-science-puzzle-explained.html

Quora (2017, March 22). What Is The Best Way To Learn Machine Learning Without Taking

Any Online Courses? Retrieved October 25, 2017, from https://www.forbes.com/sites/quora/2017/03/22/what-is-the-best-way-to-learn-machine-learning-without-taking-any-online-courses/#599a5dbc5d87

Terdoslavich, W. (2016, November 3). Real Skills You’ll Need for Machine Learning for A.I.

and Machine Learning. Retrieved October 25, 2017, from https://insights.dice.com/2016/11/03/real-skills-ai-machine-learning/

What is Big Data and Why It Matters. (n.d.). Retrieved October 27, 2017, from

https://www.sas.com/en_us/insights/big-data/what-is-big-data.html