Learn how Cloudera provides a unified platform that breaks down data silos commonly seen in organizations. By unifying the data needed for applied machine learning, organizations are better equipped to gather valuable insights from their data.
Thank you, Joao.
Our next speaker comes to us all the way from Brooklyn, New York, where the Fast Forward Labs team is headquartered. Cloudera acquired Fast Forward Labs last year to advance machine learning in the enterprise. The FFL team has a clear vision of the future and a deep expertise in applying machine learning and AI to practical business problems.
Brian Goral guides large client programs and provides operations direction for Cloudera's Fast Forward Labs team — when he’s available at the FFL offices in Brooklyn he joins the research and advising teams, bringing emerging machine learning concepts to life in client business use cases. Prior to joining the Fast Forward Labs team Brian lived a 15-plus year career focused on data collection systems and application of global data to decision-making and policy-setting. An alumnus of Michigan State University with a Masters from the University of North Carolina, Brian hails from Milwaukee though now hangs his hat in New York City. Without further adieu, please welcome Brian Goral.
Zebra Medical Vision, an Israeli startup, uses deep learning to diagnose diseases of the bone system, liver, lungs, cardiovascular system, and the brain.
Hospitals have used machine learning to predict rehospitalization for years. With new techniques, they are able to markedly improve the predictive power of readmission models.
-- Mining text from provider notes and other documents in the EHR system
-- Building models specific to individual diseases and diagnoses
CHS transformed readmissions modeling into priority scores available to care managers.
in a year and a half to a two-year period Carolinas Health System was able to drop the readmission rate from 21 percent to 14 percent.
For the insurance industry, researchers at Purdue University developed a system that uses machine learning to assess disaster damage. This makes it possible for insurers to rapidly predict claims, and serves as an independent check on human assessors.
Lloyds Banking Group uses deep learning to develop unique identifiers for each customer’s voice. The bank uses these voice profiles to confirm the identity of people who contact the call center, reducing fraud and improving operations.
Many organizations STRUGGLE to profit from machine learning
It’s difficult because it’s often hard to ask the right questions, difficult because it’s not straightforward programming - it really is science and experimentation, difficult because of the rapidly growing volume, difficult because the metrics are generally measured somewhere relative to chance which can be difficult to communicate - when a data set tells you you have a 95% certainty in a particular outcome that’s one thing, but what about when the data only supports a solution with a 65% accuracy and you need to dig deeper. Executives in a lot of organizations are not used to being given that kind of response.
You also know, better than most organizations, that a company, one of your clients, can’t outsource understanding of your own workplace or necessarily outsource their internal data product development without risking poor integration
And with so much going on under the umbrella of Artificial Intelligence these days Executives need trusted advisors to navigate a fast-moving landscape of machine learning or “AI” as many people term it.
Data products underpin much of the current business and government decision-making occurring today. Data - machine learning in particular is a huge - but difficult to execute on - opportunity for every organization.
Data products themselves are difficult at the tactical level
but there are also these strategic barriers to transformation.
Nobody knows what to buy from whom
Skills gap
The technology is moving so fast…
Departmental purchases
Different ML tools for different ML use cases
Absence of common best practices and standards
Provisioning is all over the map
One team might use Databricks another uses a competing black box software you’ll never get your data back from.
It’s important to consolidate
It saves money and process and smooths interactions - and with ML in particular there’s an added benefit - multi-task learning.
Multi-task learning is building a ML model on differently trained tasks to the benefit/enhancement of each.
In the health care world this might look something like training separate models on claims data and patient care data and finding new cost savings insights in each when the commonalities and differences in the models are combined and exploited
We hear a lot about data science “heroes” -- genius individual contributors who, singlehandedly, produce brilliant insights that change lives,
In fact, successful data science requires a collaborative approach from many contributors
Just saying something is a deep learning problem is like saying you’re going to deliver a vehicle with a 12-cylinder engine. Each of these vehicles has a 12-cylinder engine, but there are a lot of differences in producing them.
We’re there because academic research is doing some amazing work, but they’re not trying to solve your applied problem. The general formulation of an algorithm does not equal your use case solution. I can pull up some very cool use cases of deep learning using Cloudera in areas like disaster recovery, cardiac monitoring in health care, and voice identification in banking - but there is a big difference between saying that you have a deep learning challenge and bringing you to a solution. We’re here to help bridge that gap.
But obviously it takes more than good people and processes. You need the right technology.
Let’s get down to brass tacks on what the software is about
We’re based on an open source core. A complete, integrated enterprise platform leveraging open source
HOSS business model - core set of platform capabilities – we contribute actively into that community.
and we layer value added software on top - that’s how we run our business.
But what’s truly differentiating about our platform is the enterprise experience you get. It’s why we’re able to claim 7 of the top ten banks and 9 of the top ten telcos are our customers. For regulated industries, the enterprise experience is critical.
Multi-cloud – No vendor lock in. Work in the environment of your choice. Better pricing leverage
Managed TCO – Multiple pricing and deployment options
Integrated – Integrated components with shared metadata, security and operations
Secure - Protect sensitive data from unauthorized access – encryption, key management
Compliance – Full auditing and visibility
Governance – Ensure data veracity