How AutoAI Accelerates AI Adoption

Speed to market is critical for businesses to be successful.
The growing importance of artificial intelligence (AI) as a factor in decision
making means the faster a business can deploy AI for a specific task, the
faster it can reap the benefits. Therein lies the challenge. In most
businesses, data scientists must do all the heavy lifting. They must ensure the
right data is selected and in the proper format to train models. They must pick
which models and algorithms to use. And then they must fine-tune those efforts
to ensure the model delivers the most accurate predictions.

Trout out the adage necessity is the mother of invention.
Help, in the form of AutoAI
from IBM, is now available that allows non-data scientists to do some or
all this work on their own, as well as collaboratively work with the data
scientists in developing optimal models for the business.

See also: Businesses Still Struggling With AI Deployment

With AutoAI, businesses
can build and deploy a machine learning model with sophisticated training
features and no coding. The tool does most of the work for you. The AutoAI graphical tool in Watson
Studio automatically analyzes a company’s data and generates candidate model
pipelines customized for the company’s predictive modeling
problem. AutoAI automates:

• Data preparation
• Model development
• Feature engineering
• Hyper-parameter optimization

Let’s
explore these areas in greater detail.

Data
pre-processing: Most data sets contain different data formats and missing
values, but standard machine learning algorithms work with numbers and no
missing values. AutoAI applies various algorithms, or estimators, to analyze,
clean, and prepare your raw data for machine learning. It automatically detects
and categorizes features based on data type, such as categorical or numerical.
Depending on the categorization, it uses hyper-parameter optimization to
determine the best combination of strategies for missing value imputation,
feature encoding, and feature scaling for your data.

Automated
model selection: The next step is automated model selection that matches
your data. AutoAI uses a novel approach that enables testing and ranking
candidate algorithms against small subsets of the data, gradually increasing
the size of the subset for the most promising algorithms to arrive at the best
match. This approach saves time without sacrificing performance.  It enables the ranking of many candidate
algorithms and selecting the best match for the data.

Automated
feature engineering: Feature engineering attempts to transform the raw data
into the combination of features that best represents the problem to achieve
the most accurate prediction. AutoAI uses a novel approach that explores
various feature construction choices in a structured, non-exhaustive manner
while progressively maximizing model accuracy using reinforcement learning.
This results in an optimized sequence of transformations for the data that
best match the algorithms of the model selection step.

Hyperparameter
optimization: Finally,
a hyper-parameter optimization step refines the best performing model
pipelines. AutoAI uses a novel hyper-parameter optimization algorithm optimized
for costly function evaluations such as model training and scoring that are
typical in machine learning. This approach enables fast convergence to a good
solution despite long evaluation times of each iteration.

Summary

An
article by Jacques Roy, worldwide Digital Technical Engagement lead on
Watson Studio and Watson Machine Learning, at IBM, put the benefits of AutoAI
into perspective. In that article, he noted:

“Overall, AutoAI democratizes data science and AI —
data preparation, model development and selection, execution, and deployment.
This addresses the shortage of data scientists and gets to a solution faster.
By accelerating the data science lifecycle with AutoAI, businesses can focus
more on high value-added work and innovative solutions.”

The
bottom line is that AutoAI offers:

• Faster model selection: Users select top-performing models in only minutes.
• Faster starts: Users can quickly get started with experimentation,
  evaluation, and deployment.
• AI lifecycle management: AutoAI enforces consistency and
  repeatability of end-to-end ML and AI development.