Students will submit a research paper which will provide a survey of machine learning algorithms and their application to different research areas. The paper must include a critical review of current...

Student Name
Course Name
University Name
05
th
-May-2017
Introduction
Machine learning has delivered big technology breakthroughs recently and is regarded as the
major driver toward artificial intelligence. It will greatly impact most industries over the next
three to five years, but data and analytics leaders should act now to understand its benefits and
challenges.
Within a few years, deep learning will greatly influence analytics tools and practices. Its effect
has already been felt by smartphone users: deep learning changed speech recognition with
Apple's Siri and Google's Voice from a curiosity to a usable feature. There are almost weekly
breakthroughs in image recognition, speech-to-text conversion, machine translation, and in many
more areas.
Gartner estimates that over 2,000 vendors, from startups to tech giants, are launching deep-
learning tools, cloud services, APIs, packaged applications and consulting services. This creates
hundreds of deep-learning-based options for organizations of almost all industries. Data and
analytics leaders have a unique opportunity to assess, in very practical terms, how deep learning
can benefit their digital business.
Definition
Machine learning is a variation on machine learning: business problems are solved through the
extraction of knowledge from data. Machine learning expands standard machine learning by
allowing intermediate representations to be discovered. These intermediate representations allow
more complex problems to be tackled and others to be potentially solved with higher accuracy,
fewer observations and less cumbersome manual fine-tuning.
Description
With deep learning, a computer model can be fed lots of complex data, such as images, speech
and text. For example, deep-learning algorithms can analyze retina scans to "figure out" on their
own which patterns indicate healthy or diseased retinas (and indicate the specific disease). The
"figuring out" process relies on brute-force, high-performance computing and can, to some
extent, render obsolete the tedious handcrafting of features and data preparation.
Benefits and Uses
Machine learning inherits all the benefits of machine learning (Table 1). In addition, deep
learning's biggest promise is to learn domain-specific intermediate representations, which boosts
the performance of resulting solutions. Several recent breakthroughs in cognitive domains (e.g.,
image recognition, machine translations, speech recognition) demonstrate this:
 Image recognition — In late 2015 and early 2016, Microsoft's Machine Residual
Networks (ResNet) and Google's GoogLeNet (v4) showed stunning image recognition
systems that surpassed the performance of humans for the ImageNet image classification
task.
 Machine translation — Google has unveiled Google Neural Machine Translation
(GNMT) and claims significant improvement over past state-of-the art, in-machine
translation.
 Speech recognition — Baidu's speech-to-text services outperform humans at similar
tasks.
Further exciting developments can be seen in noncognitive domains, where deep learning is
making inroads:
 Fraud detection — PayPal is using deep learning as a best-in-class approach to block
fraudulent payments.
 Recommender systems — Amazon has applied deep learning for best-in-class product
recommendations.
Most organizations lack the necessary data science skills for even simple machine-learning
solutions, let alone for deep learning. Analytics leaders should assess their organization's
business requirements in light of the expected high costs of developing custom deep-learning
solutions.
Adopters of deep learning are likely to fall into one of four groups:
1. Cutting edge — An estimated 50 to 100 organizations, all of them vendors, can produce
or...