Criteo

Criteo is the advertising platform for the open Internet, delivering effective advertising across all channels, by applying advanced machine learning to unparalleled data sets.

What advertisers really need to know about Deep Learning

  • Colin Barnard, Commercial Director at Criteo, ANZ
View all images

By Colin Barnard, Commercial Director at Criteo, ANZ

Deep Learning is often perceived as the solution to almost all problems in digital advertising. As the youngest child of the AI cosmos, Deep Learning promises to increase relevance, improve prediction and reduce banner blindness.

Deep Learning isn’t a child anymore. Researchers have studied Deep Learning for more than 20 years with significant developments in image recognition and text and sound processing. But how do these new developments relate to digital advertising? Especially considering ads are implemented in real-time and utilise more complex data than standard bare pixels and sound frames.

The Swiss Army knife of digital advertising

You could say Deep Learning is regarded as the Swiss Army knife of digital advertising. The Swiss Army knife is a multi-purpose tool but to solve more complex scenarios, we often need to use a Swiss Army knife in conjunction with other tools or source the help of a handyman. 

This notion can be applied to Deep Learning. Deep Learning may be the most powerful subset of Machine Learning, perfectly suited to solve tasks like image recognition, but for digital advertising where large amounts of quality data and years of experience are required, this all-purpose tool shouldn’t be the only ace up a marketer’s sleeve.

Machine vs Deep Learning?

It’s important to understand how Deep Learning fits into the Machine Learning spectrum:

  • Supervised Machine Learning teaches algorithms to view data and cluster them respectively to make predictions. An everyday example of this is the spam filter in our inbox. In digital advertising, Deep Learning can predict the likeliness of a user clicking on a banner for example. It involves defining the features to generate a label as an output, like “This mail is spam” or “This user has a Click Through Rate of X”.
  • Unsupervised Machine Learning finds patterns in a large pool of data, analysing the calculations’ results to classify consumer behaviour. You don’t have to define a feature or label, instead the machine looks for interpretable clusters of patterns.
  • Deep Learning is a subset of Machine Learning. Based on artificial neural networks, Deep Learning can be supervised, semi-supervised or unsupervised using high-speed machines that have computed, analysed and stored huge amounts of data in the past. But there’s a catch.

Should advertisers use Deep Learning? The answer is just a little.  

Mastering Machine Learning

The challenge with Deep Learning is that it requires an enormous amount of data to train such a complex system and for advertisers this means processing the data in real-time.

Advertising is driven by programmatic-buying technologies under stricter latency constraints than other Deep Learning practices (single-digit milliseconds at most), so the processing of this data can only be achieved with a major increase in computer power, but this is only justifiable by massive uplifts that are almost impossible to achieve.

You’ll notice that Deep Learning is rarely used in bidding stages and instead for precomputing features outside the critical path. A lot of advertisers don’t realise that these features can be fed into a much simpler, traditional Machine Learning model or a logistic regression model with higher speeds to generate a better result.

A logistic regression model is a single layer model that processes features that are usually hand-crafted and is often used as the last layer of a Deep Learning model. If you have a good feature list and enough data, logistic regression provides a faster solution with less power than Deep Learning.

New technologies are often labelled as revolutionising the advertising industry but what advertisers must do is distinguish between potential game changers and marketing jargon. It’s important for advertisers to not buy into the hype of Machine Learning verses Deep Learning or rely on one single tool but rather determine whether the technology contributes to their objectives.

Deep Learning is becoming central to marketing strategies but only in the wider context of Machine Learning, including tree-based and regression models and self-organising AI networks. For the best impact, advertisers must take a scientific approach to experimenting on their data and KPIs while using an optimum combination of Machine Learning tools including Deep Learning. Advertising platforms like Criteo can help guide advertisers through this process to create highly personalised customer experiences. 

For more information on how to utilise Machine Learning tools including Deep Learning, visit www.criteo.com.

Colin Barnard, Commercial Director at Criteo, ANZCredit: Colin Barnard
Colin Barnard, Commercial Director at Criteo, ANZ

Colin Barnard is Criteo’s Commercial Director ANZ. Colin is a proven business leader with over 16 years of experience in digital marketing. Prior to Criteo, Colin was Head of Retail and Google Shopping ANZ at Google. He has experience in both the UK and AU/NZ markets, and is passionate about helping businesses unlock the power of technology to make their marketing targeted, efficient, relevant and accountable.




Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.
Show Comments
cmo-xs-promo

Latest Videos

More Videos

Thanks for your feedback, Rabi. While we introduced the ROO concept using a marketing example, I also believe that it is pertinent to man...

Iggy Pintado

Introducing Return on Outcome (ROO) - Brand science - CMO Australia

Read more

Thanks for your insight, Philip. Return On Outcome (ROO) requires balanced thinking with the focus on outcomes as opposed to returns.

Iggy Pintado

Introducing Return on Outcome (ROO) - Brand science - CMO Australia

Read more

Beautiful article.

Hodlbaba

15 brands jumping into NFTs

Read more

"Blue" is really gorgeous and perfectly imitates a human customer support operator. Personally, I won't order a chatbot development for m...

Nate Ginsburg

Why the newest member of BT’s contact centre is a chatbot

Read more

As today’s market changes rapidly, the tools we use change, and it is important to adapt to those changes to continue to succeed in busin...

Anna Duda

Report: 10 digital commerce trends here to stay

Read more

Blog Posts

How the pandemic revealed the antidote to marketing’s image problem

What does marketing truly ‘own’ in most organisations? Brand and campaigns, definitely. Customer experience? That remains contested ground.

Murray Howe

Founder, The Markitects

Still pursuing a 360-degree view of the customer?

On the Internet, nobody knows you’re a dog.” It may have been true in 1993 when this caption to a Peter Steiner cartoon appeared in the New Yorker. But after 30 years online, it’s no longer the case.

Agility in 2022

Only the agile will survive and thrive in this environment and that’s why in 2022, agility will need to be a whole-business priority.

Sam McConnell

Melbourne bureau chief, Alpha Digital

Sign in