Page 17 - AeM_August_2017

RESEARCH, ANALYSIS & TRENDS

sion making. In other words, it is how humans learn With deep learning however, there are no pre assumed
from practice as we try many different approaches be- rules. Computers learn by practicing what will be the
fore making a final decision. best combination, should it be the next pair of black
shoes or a better option – brown sandals, matching
Thanks to that self-learn procedure, it intuits from expe- bags, etc. The main drive is that no human being has
rience how to play optimally – without adding human programmed a computer to perform any of the specific
input or manually implemented rules. actions described above – every display is driven by
data and algorithmic learning. In other words, human
Deep learning makes computers ultra- action is only required at the step of programming the
algorithm how to program itself.
precise

Deep learning is one of the most promising subfields of What will the future bring?
AI research. Today, the technology is finding many ap-
plications across any industry that relies on technology. A study on “Ad Spending in Southeast Asia” by eMar-
keter and IAB Singapore revealed that digital ad spend-
Ultra-precision offered by deep learning has been also ing share of companies in the region will almost double
used in the advertising industry. These self-learning up to 53.5% by 2020. This growth will be driven by in-
algorithms offer highly accurate product recommenda- creasing demand for mobile, video, and social media
tions, while also better predict the probability that a user advertising. As consumers seek better experiences
clicks on an ad (conversion potential) or the value of from brands, regardless from offline or online service
purchase (conversion value). This will allow more effi- providers, personalisation will be a key differentiator for
cient advertising activities for brands. companies in Southeast Asia.
Deep learning will play an increasingly important role
Deep learning in advertising illustrates the use of algo- not only in the advertising industry, but also across dif-
rithms on a massive amount of data and act like hu- ferent sectors. As the volume of data is set to surge,
mans without specific instructions or rules. It can work companies should consider implementing these new AI
inter alia in recommendations used not only by e- technologies to analyse results and help understand
commerce companies to persuade customers to buy our world better, as the game has just begun. ◊
additional products, but also by other companies to sug-
gest music, events or even dating profiles. By Daniel Surmacz,
COO, RTB House
The typical approach to targeted ads (personalised re-
targeting) is as follows: A user sees a banner creative
based on a matter of predefined assumptions - if you
checked out black shoes with a gold clasp, the recom-
mendations might want to show black boots with buck-
les from historical information gathered.

MediaBUZZ Pte Ltd - Independant ePublisher for Asia