Never Done Learning: How I Became a Machine Learning Enthusiast
Summary: Encouraged by Shawn Rutledge, Chief Scientist at Visible Technologies, several engineers within the company participated in Stanford’s free online Machine Learning course. Lucas Parker, Lead Software Development Engineer at Visible, relates his experiences as a newcomer to machine learning, comparing them to the challenges of social media monitoring. Additionally, both Shawn and Lucas take part in an online machine learning competition, showing that experience is a significant factor in the effectiveness of building and applying the sorts of analytical models that bring value to Visible’s customers.
It was September 9th of this year that Shawn Rutledge, Chief Scientist and head of the Core Technology team at Visible, gave the clarion call to our engineering team to participate in Stanford’s forthcoming experiment in online learning – namely the Machine Learning class taught by Professor Andrew Ng. A few of us heard the call and rushed into a learning experience that drew us into a discipline outside the typical realm of software engineering. Shawn organized the email discussion group and helped to answer questions along the way as we all became neophytes in the esoteric field that helps to differentiate Visible Technologies in the budding industry of social media monitoring and analytics.
Machine learning is the science and sometimes black art of leveraging computational power to extract information out of raw data. Statistical models are built with as much understanding of the context of the data as is reasonable in order to attempt to accurately classify or make predictions about future data. Nowhere does this seem more relevant to me than in the milieu of social media, where the data are noisy and often difficult to interpret; where it is conceivable that a solitary, short, grammatically incomprehensible blurb may contain a greater amount of pertinent information than any ten voluminous, loquacious, ostentatious essays on the same subject. The desire to expand on old ways or find new ways to obtain value from data is a constant here and seems eventually to work its way into everything we do as an organization, so naturally this was an opportunity that could not be missed.
To say that I am not a mathematician is an understatement. I am a software engineer, yes, but there are many ways to approach problem-solving, some more formal and others more mechanical. My natural tendencies are toward the latter, so this exercise in formal understanding was an eye-opener for me. The class was not focused on theory, but that it favored applied machine learning over an exhaustive examination of the mathematical foundations brought it together for me. I love to get my hands dirty.
It was six or seven weeks into the class when I stumbled over Kaggle , a website that hosts machine learning and statistical modeling competitions for private interests. Here was a treasure trove of opportunities to apply my newfound status as a novice machine learning practitioner. So I entered the Don’t Get Kicked competition. I won’t delve into the technical details, but suffice it to say that if your current understanding of machine learning is what mine was prior to this last September, the problem space would seem impossible to navigate. As it is, the problems are merely exceedingly challenging. I’ve spent my last four weekends and more than a dozen weeknights working almost non-stop, attempting to hammer out algorithms to build models to gain on my competition. My goal was (and as of this writing still is) 50th place on the leaderboard, a seemingly modest goal but one that seems ever further away as I conduct experiment after countless experiment, working feverishly to try to eke out those next few little points of performance.
Meanwhile, latching onto my excitement, Shawn himself entered the competition and with what for all appearances was an entirely blasé attempt, managed to make it into the top 20 so fast it would make your head spin. I can imagine him scratching his lapel and having a good yawn over it right now.
There are several pearls of wisdom in all of this for me. First is that turning data into information is hard. When I look at the value of the information that we at Visible present to our customers, extracted from raw social media data and made available in so many useful ways, I am all the more astounded for my fresh experience as an amateur in the field. Second, experience counts. The staggering distance between the results of my hundred-and-change hours and Shawn’s two or three is a very clear indicator that experience can mean the difference between merely effective and impressive. Third is that keeping people learning is by far the best way to capitalize on human resources. Nothing promotes new ideas, engagement and dedication like the drive to learn something new and to apply it to make work and life a little better for everybody.
The result of all of this is that I feel very fortunate to be in the position that I’m in at a company that takes on extraordinary challenges and brings incredible value while promoting learning within the organization to challenge all of us to excel in many dimensions. I am also humbled by the nature of the challenges we face, and such challenges are a strong motivator for me.
Machine learning has become an overnight passion for me that I intend to continue honing. You can find me on Kaggle here. Several of we engineers will also be taking part in the free online Natural Language Processing course being offered by Stanford starting in January. Come find us there.
Communications Best Practices
Get the latest updates on PR, communications and marketing best practices.
Cision Product News
Keep up with everything Cision. Check here for the most current product news.
Thought leadership and communications strategy for the C-suite written by the C-suite.
A blog for and about the media featuring trends, tips, tools, media moves and more.