Why You Should Get Into Machine Learning Now – Kris Bolton Scroll to top

Why You Should Get Into Machine Learning Now


Kris Bolton - July 26, 2018 - 0 comments

Chapter Two: Why You Should Get Into Machine Learning Now

This post is formed of the entire of Chapter Two from Practical Deep Learning Fundamentals, one of the four chapters which are freely available. This chapter is intended to inspire the reader to take an interest in machine learning. You can read chapter one and two in draft PDF form by subscribing to email updates. Chapter three and a technical chapter, likely chapter 5 will also be available via email updates over the next two weeks.

Growth

Since 2006 the machine learning field has been growing at an unprecedented pace. New developments each year push growth upwards and it shows no signs of slowing. Machine Learning has capabilities unheard of in computer science, from the outside it seems very close to human level intellect. Sadly, this is not the case.

However, machine learning techniques and methods are some of the most powerful created and provided the new state-of-the-art in most of the areas they are applied. This has further increased the growth as academics, individuals and businesses hear of the abilities of these techniques they want to apply them to new problems.

Google Trends can provide anecdotal evidence of the growth of machine learning. Below is a graph produced by Google Trends using Machine Learning as a search term plotting the interest globally over the last five years.

Further anecdotal evidence of growth can be seen analysing specific machine learning tools and libraries using Google Trends, “TensorFlow”, “learn TensorFlow”, “Scikit learn”, etc.

Mergers and Acquisitions

The value of machine learning has not been lost on major corporations in the technology sector. Google, Facebook, Apple have acquired and absorbed business focusing on machine learning into their various areas of their business to add functionality to their products.

The timeline above shows the acquisitions of Google, Apple and Facebook since 2010. At first glance you can see the increased speed of acquisitions. Upon closer inspection, you will notice acquisitions relating to existing services, or new services which launched soon after the acquisition. Mergers and acquisitions are often shrouded in secrecy to protect commercially sensitive information, so it can be difficult to gain much insight into the transactions, however researching the companies listed can provide some interesting information. Others are more obvious.

Siri, acquired by Apple in 2010, became the core of Apple’s mobile virtual assistant. Google has made a number of acquisitions to further its own virtual assistant, Banter and API.ai provide conversational AI and advanced natural language processing.

Career

As businesses increase the size of their machine learning workforce; forming new machine learning projects and attempt to keep up with competitors, an opportunity presents itself for the reader. According to research conducted by LinkedIn, Machine Learning positions occupy the top spots for the fastest growing jobs on LinkedIn.

LinkedIn monitored the growth of users adding their current position in various fields and compared the results over five years (2013-2017) and compiled an ‘Emerging Jobs Report 2017’. The following are machine learning specific.

  1. Machine Learning Engineer 9.8X growth.
  2. Data Scientist 6.5X growth.
  3. Big Data Developer 5.5X growth.
  4. Director of Data Science 4.9X growth.

Four of the top ten, and the two highest growth jobs in the report are machine learning related. Incidentally, another three are within the computing field making computing easily the fastest growing job sector on LinkedIn.

Earning potential should also be considered. Salaries will not be discussed here, as they are highly dependant on location. However, the job intelligence websites Payscale and Glassdoor can provide an indication, or detailed stats on the earning potential of various jobs.

Resources

Hardware

One of the largest historical obstacles to machine learning progressing as a field was the limited availability and power of computing hardware. This is no longer the obstacle it used to be; the growth and advancement of computer hardware, specifically graphical processing units (GPUs), and the advent and growth of infrastructure as a service (IaaS) and platform as a service (PaaS).

An unexpected accelerator of the computer hardware industry has been the cryptocurrency industry. In early 2018 enthusiastic crypto miners looking to make high profits created a global shortage of GPUs, causing some retailers to limit the number of cards individuals could purchase. During this GPU card shortage, industrial crypto miners started purchasing the processor wafer, the material CPUs/GPUs are made of, to create their own custom cards. In January 2017 the manufacturer TSMC were selling more 16nm wafer to crypto miners than nVidia.

Ignoring what this means for the average consumer, I would argue the increased demand and revenue for card manufacturers and wafer manufacturers can only accelerate the creation of faster and more efficient hardware, which will become the backbone of future machine learning research and commercial endeavours.

Software

Powerful open source machine learning software libraries have emerged over the last five years. The most popular of these machine learning libraries have come from powerhouses of computing such as Google, Facebook and Microsoft. Written by experts in their field, tested and used on real-world problems and programs within their origin company. And they’re available for you to use.

TensorFlow is one such library. The most popular machine learning library on GitHub, created by Google’s Brain team and open sourced in 2015. A unique visual dashboard allows the visualisation of complex machine learning algorithms, and the user base numbers in the thousands and a number of informative and tutorial books have been released, so support is readily available. For these reasons, this book uses TensorFlow for a number of the projects in future chapters.

Impact

Quite simply, machine learning can have a positive impact on your life, the life of users of services you create, your country’s economy and the global economy.

Now is the time to get into machine learning

References

Google Trends: machine Learning (2018). [Accesed July 2018]. https://trends.google.com/trends/explore?date=today%205-y&q=Machine%20Learning

CB Insights: The Race for AI (2018). [Accessed July 2018]. https://www.cbinsights.com/research/top-acquirers-ai-startups-ma-timeline

LinkedIn: The Fastest Growing Jobs in the US (2018). [Accessed July 2018]. https://blog.linkedin.com/2017/december/7/the-fastest-growing-jobs-in-the-u-s-based-on-linkedin-data

LinkedIn: Emerging US Jobs Report 2017 (2018). [Accessed July 2018]. https://economicgraph.linkedin.com/research/LinkedIns-2017-US-Emerging-Jobs-Report

PC GAMES N: TSMC are currently selling more 16nm chips to cryptocurrency miners than Nvidia (2018). [Accessed July 2018]. https://www.pcgamesn.com/tsmc-bitcoin-supply-nvidia

Dark Vision Hardware: Bitmain is buying 20k 16nm wafers from TSMC per month (2018). [Accessed July 2018]. https://www.dvhardware.net/article68109.html

 


Practical Deep Learning Fundamentals will be released on September 25th 2018.

Related posts

Post a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.