Machine Learning

The STEM field, which is an abbreviation for science, technology, engineering, and mathematics, can be very broad when it comes to the different roles related to it. (Sometimes it is also referred as STEAM by which the "A" stands for arts.) And yet can we really say that we have a role in only one of the letters of STEM? This abbreviation was made for the simple reason that they all go along together and can practically be interchangeable in some occasions. One example of a role in the STEM field is Machine Learning. So lets go over what is and does a Machine Learning developer.

What is Machine Learning?

"Machine learning is a branch of artificial intelligence (AI) and computer science which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy." - https://www.ibm.com/cloud/learn/machine-learning

Machine Learning is a sub-branch of computer science and artificial intelligence that, although one might think that is solely related to actual machines, tends to focus on the study of computer algorithms and how these have the capacity of improving by using data and previous experience. This use of statistical methods allow algorithms to be trained into making decisions by either making classifications or predictions and use this new data (what the machine learned) so we can make our own predictions and learning (based on our interpretations).

Lets take for example a chess game. We play it around on our computers and every time we try to play with the highest difficulty we cant seem to win against the computer. Well how does the computer actually even is capable of playing chess with us anyways? Using data from previous games of chess played by two human players we can set up an algorithm that will practically analyze each move and learn from it. By learning from this data that is fed to the algorithm now the computer is capable of making decisions on the chess game based on previous data and experience and use it to predict what would be the best move and what your next move is going to be.

Why is Machine Learning important in STEM?

"The nearly limitless quantity of available data, affordable data storage, and the growth of less expensive and more powerful processing has propelled the growth of machine learning. Now many industries are developing more robust machine learning models capable of analyzing bigger and more complex data while delivering faster, more accurate results on vast scales." - https://quantilus.com/why-is-machine-learning-important-and-how-will-it-impact-business/

With this excerpt we are able to appreciate how machine learning is not simply a new field or specialization within STEM but it is also a very important tool or aid for many other fields be it for managing data, predicting or learning, or simply by using various algorithms to bring up a better, faster and efficient experience which allows for more progress int other STEM fields.

What makes this position interesting and unique?

"Machine learning is one very exceptional technology that uses algorithms and statistical methods to perform its functions without needing human intervention. This kind of technology gives machines a brain such that they can automatically readjust themselves to suit their purpose. For instance, a piece of automated factory machinery recalibrates its settings if it senses an abnormal temperature rise. To counter the heat, it automatically opens the machine’s vents and power up the fans to allow cool air to come in. And if the issue remains unsolved during this operation, it sends out an alert to a technician on duty. With such technology by our side, anything is possible." - https://www.westagilelabs.com/blog/pros-and-cons-of-implementing-machine-learning-in-your-projects/

The possibilities of machine learning are endless and we should not limit ourselves in the one track though that robots are going to take over. If we take into account the type of tasks or jobs that we are able to make automatically which allows us more time, safety, and efficiency then the possibilities are endless. Taking into account not just the business or science aspect but also the human aspect we can use machine learning for tasks that can normally be dangerous or that they can pose harm or threat to humans, whether it is immediate or in the long run.

What makes this role similar to others?

The similarity that can be held when it comes to Machine Learning and other fields is the very fact of how these are intertwined with one another constantly. We cant really say that all fields work seperately since they complement each other for further growth. We can see techniques, programming languages, tools, algorithms, and other elements that can be used in different fields yet implemented differently.

What specific programming languages and tools could one expect to work with in this position?

Like many other STEM specializations machine learning has certain specific programming languages and tools which are best for this type of job. Luckily it doesn't have to be limited to only one or two programming languages since it can be very flexible or versatile with the different ways to work around it. Some of the programming languages that can be used to work on machine language are: Python, R, JavaScript, Java, Julia, Lisp, Scala, C/C++, Typescript, GO, and Shell.

We do need to take into account that by using a programming language alone will not suffice for what machine learning really is since many of these programming languages need support in the sense of tools, frameworks, libraries, among others. So of the common tools used for support when it comes to working around machine learning are:

• PyTorch: Autograd Module, Optim Module, nn Module
• Apache Mahout: Preprocessors, Regression, Clustering, Recommenders, Distributed Linear Algebra.
• Shogun: Regression, Classification, Clustering, Support vector machines, Dimensionality reduction , Online learning etc.

What is an example of a problem or a challenge someone in this role could solve or be asked to work on?

Lets take into account for an example the predictability of machine learning, With proper data introduced constantly machine learning is capable of improving the chances of predicting outbreaks and epidemics that can occur making work more efficient for a epidemiologist. On the other hand it can also aid the meteorologist in predicting anomalies in the weather. These are common challenges that can be tackled by someone from machine learning specialization that normally would be considered a mystery or uncertain to know.

What are some positives and negatives about this position?

As any and every other position in STEM, as well as any other field that is not included in STEM, we always have to take into consideration the positive and negative aspects of it. The pros and cons you may say. Some of the good and bad things of machine learning that we can take into account could be:

Pros:

• Identifies trends and patterns very easily: A positive aspect of machine learning is its ability to identify trends and patterns which will decrease the amount of human interaction and therefore speed up a task or process.
• It improves itself over time: The capacity of machine learning allows for it to adapt and continue learning. It readjusts when necessary and because of it its capable of predicting any outcome based on previous experience and adaptations.
• It is self-sufficient and assorted: By assorted, in this case, one might mean on how it is able to be used in any kind of field be it healthcare or industrial machinery or many others.
• Saves time and is energy-efficient: This STEM specialization is able to reduce the amount of interruptions needed to give any type of maintenance or needs of calibration since it can readjust itself if necessary. Hence, by doing so the reduced number of repairs or shutdowns are able to save time and be energy efficient.

Cons:

• Errors are frequent and take a long time: An error that can commonly occur when you are a developer is to have any sort of error on the code, be it syntax, logical, arithmetic or compilation errors. Thus the fixing of any error on the algorithm or code for any type of machine learning can be extensive and can take time.
• It is expensive: As mentioned previously errors can take a lot of time to fix and if it needs time then it costs money. It can imply many costs in labor and testing to have it functioning. Not only during the fixing process but also in the development process machine learning can be quite expensive.
• Has to be specialized for every project: This one can be kind of like a no brainer but each project has its specifications and different elements that need to be implemented. There are workarounds to this such as polymorphism and heredity when it comes to coding but even so it can be rather difficult and it can cause the previous two cons to constantly resurface.

When it comes to any field in STEM no blog would suffice to give even a proper explanation of it. With everything that I just wrote I am certain that I am just covering, barely, the tip of the iceberg or what machine learning is. I hope this can motivate someone or just anyone to pursue their interests if this may be it. As always I take no credit for any images.