Machine Learning’s Business Impact

CEO Robbie Allen recently spoke about the business impact of machine learning at Adobe’s Think Tank on The Future of AI in the Enterprise.

“I really focus on three areas where I think that machine learning specifically will have an impact,” he said.

“One is reducing costs, and by that I don’t mean that we’re displacing the workforce. What I mean by that is that there’s lot of jobs out there that were never intended for people to do. But the only option was for people to do them, because we didn’t have software that could it on their behalf. So I think there will be a lot of those types of jobs – repetitive – it’s not ideally suited for humans to do in the first place.”

“Second is increasing efficiency, and oftentimes you’ll hear about potentially given people superpowers in their job, where not only can they do their job, they can do it a lot faster, a lot better, and they can be much more productive.”

“And then the third are is achieving new breakthroughs, and this is something I think is going to be really powerful as it relates to what you can get out of machine learning. As a society we have blindspots that we’re not aware of. For any new breakthrough that occurs, that only happened because someone else didn’t think to do that. Machine learning gives us an opportunity to discover new patterns, new ways of working that weren’t obvious before, and so I think we’re going to be able to see all sorts of interesting new capabilities in the enterprise as it relates to that.”

During the talk, Robbie was clear about the need to focus squarely on solving real business problems:

“So I think especially  when it comes to the enterprise, it’s less about ‘can we properly define AI,’ and really what problems are we trying to solve,” he said.

“And it goes back to, what are the key differences now between what we can do now versus some of the vendors that we’ve seen before. And it’s the computing power, it’s the availability of data, and it’s the advanced algorithms that have been created.”

“And I think really focusing on what problems can you solve with that, that’s really where it’s at for the enterprise, much more so than ‘are we ever going to create AGI, artificial general intelligence? Are we going to have sentient beings?’ There’s no path for that in the research right now and so to me it’s not even really worth talking about.”

For Machine Learning, Data Comes First

CEO Robbie Allen recently spoke about the importance of data strategy at Adobe’s Think Tank on The Future of AI in the Enterprise.

“People hear about artificial intelligence and all this crazy stuff that it’s doing and so they think that it’s going to be this magical solution that solves all of their problems,” he said.

“In most cases, in my experience, it’s better to walk before you can run when it comes to this, especially if you’ve never done a machine learning style project before which requires a significant amount of data. You first have to to get your data strategy in place first before you can have an AI strategy.”

Later, Robbie spoke about a common data issue many enterprises may not be expecting: the high cost of getting data out of their systems.

“It’s likely that companies will spend just as much money on getting data out of their systems as they did to implement the systems to begin with,” he said.

“What I mean by that are ERPs, CRMs, all their databases. To get the data out of it in a format that’s useful for machine learning algorithms and other things is a nontrivial task.”

“And it turns out to be, in most cases, the long pole in the tent. That is, the hardest aspect of actually implementing a project is just to get the data. So that’s why we were talking about before, data strategy, when do you get it – it’s going to be difficult to do much in the space in the enterprise unless you have access to data in a format that makes it available to somebody to use in terms of implementing algorithms.”

 

Our 3D Approach to Machine Learning

 

Excerpt from an interview in Adobe’s “Glass Tank” at The AI Conference in New York City, May 1, 2018.

—

Infinia ML is a team of machine learning experts focused on making real business impact. To deliver that impact, we have a wholistic way of helping companies achieve their goals.

We call it our 3D approach:

Data Preparation

We begin by assessing whether your data is ready for machine learning. Data readiness, not algorithms, is often the biggest obstacle to machine learning success.

We’ll make sure your data is accessible, sizable, useable, understandable, and maintainable before we give you the green light.

Development

With your data prepared, we apply our scientific knowledge of machine learning to your business challenges.

We design algorithms, build models, test results, and show you what’s possible.

Deployment

Algorithms that don’t work in the real world don’t drive real results for your business.

Our team integrates our solutions with your systems, helping your organization achieve the true impact of machine learning.

—

Whether you need help with just one D or all of the above, we’re ready to talk. Contact us to learn more.

Catch Us on Both Coasts: Our Spring Events

Bridge

We hit the road to talk about machine learning!

Your Event Could Be Next

If you’re looking for a world-class speaker on machine learning, deep learning, or artificial intelligence, we’d love to talk.

RECENT EVENTS
The Future of Operations in the Robotics Age

Wednesday, March 7th in New York City

Our Executive Chairman Mike Salvino delivered the 1:15 keynote at the FORA Summit presented by HfS Research.

In front of a group of service industry leaders and practitioners at HfS FORA,” wrote diginomica, “Salvino blasted his way through the “machine learning epidemic,” with actionable advice mixed in.”

Duke ML Day

Saturday, March 31st  in Durham, North Carolina

We were proud sponsors of Duke’s first machine learning day, and excited that our Chief Scientist Larry Carin was a speaker.  Of course, he’s also Duke’s Vice Provost for Research and a Professor of Electrical and Computer Engineering.

Our CEO Robbie Allen was on hand, too.

Triangle ML Day

Tuesday, April 3rd in Durham, North Carolina

Because one ML Day just isn’t enough, we were proud to sponsor this one too. It also took place at Duke University, and our Chief Scientist Larry Carin spoke once again. Some attendees even scored some swag from our Infinia ML booth!

Applied Artificial Intelligence Conference

Thursday April 12th in San Francisco

CEO Robbie Allen was a panelist on a session called “Savior or Destroyer? How AI Is transforming The Media.”

Webinar: A reality-check on Enterprise Artificial Intelligence (AI)

Thursday, April 12th

Executive Chairman Mike Salvino joined the panel on this webinar from HfS Research.

Carolina Family Office Forum

Tuesday, April 24th in Raleigh, NC

CEO Robbie Allen and Executive Chairman Mike Salvino speak to family office executives on the disruptive potential of artificial intelligence and machine learning.

Adobe Think Tank

Monday, April 30th in New York City

CEO Robbie Allen joined a forum on the future of AI in the enterprise.

The Artificial Intelligence Conference

Tuesday, May 1st in New York City

CEO Robbie Allen revealed “Best Practices for Machine Learning in the Enterprise.”

NC TECH 2018 State of Technology Conference

Friday, May 4th in Durham, NC

CEO Robbie Allen spoke on the past, present, and future of artificial intelligence.

Triangle CHRO Association

Friday, June 8th in Cary, NC

Executive Chairman Mike Salvino and CEO Robbie Allen speak to senior HR officers about the realities of artificial intelligence and machine learning at 8:30am.

The Copyright Society of the USA’s Annual Meeting

Sunday June 10th – Tuesday June 12th in Toronto (ironically)

CEO Robbie Allen joined the AI & Copyright panel to help legal experts understand the possibilities for natural language generation.

 

 

Mice & Machine Learning Help Map the Mind

In a new paper co-authored by Infinia ML Chief Scientist Larry Carin and published in the journal Cell, machine learning gave scientists a new way to understand and treat depressed brains. Today, mice are the subjects. Humans could be next.

DURHAM, N.C., Mar. 1, 2018 — Mice brains and machine learning may lead to a new way to treat depression, according to a new paper published in the journal Cell and co-authored by Infinia ML Chief Scientist Larry Carin, Ph.D.

The paper describes how scientists measured electrical signals in the brains of both observably resilient, active mice and observably depressed, inactive mice. The complexity and scale of the available data, gathered from 18 regions of the brain, then required advanced machine learning for analysis. In effect, scientists trained a learning algorithm to map each brain’s connections. They found a pattern in the resilient mice that differed from the depressed.

Larry Carin, Ph.D.
Infinia ML Chief Scientist Larry Carin, Ph.D.

“We wanted to understand the traffic flow of a healthy brain,” said Carin, the project’s machine learning lead. “That had not been done before, and machine learning helped us overcome that key technical challenge.”

This new understanding of the brain’s electrical system brings new potential for treatment in mice. More importantly, the research lays groundwork for future advances in human mental health. When scientists measure the relevant patterns in human brains, advanced machine learning could help them assess and treat depression.

Meanwhile, Carin’s company, Infinia ML, is already busy applying machine learning techniques to biological and medical breakthroughs from cancer detection to genetic screening.

“Machine learning offers new ways for us to understand our bodies and minds,” said Carin. “And the best part is, we’re just getting started.”

About Infinia ML

Infinia ML empowers companies to make smarter decisions and automate complex business processes by leveraging the latest breakthroughs in machine learning. Infinia ML has a team of leading AI researchers and deep learning experts that have published hundreds of peer-reviewed papers through top machine learning conferences and journals.

Backed by noted private equity firm Carrick Capital Partners, the Durham, North Carolina company is led by CEO Robbie Allen, an experienced AI entrepreneur, and Chief Scientist Lawrence Carin, Ph.D., the Duke University Vice Provost for Research and Professor of Electrical and Computer Engineering. Learn more.

Is Your Business Ready for Machine Learning?

 

Are you ready to implement advanced machine learning solutions in your organization?

Your team’s ability to lower costs, increase efficiency, and achieve new breakthroughs will be shaped by its machine learning background, your data policies, and, most importantly, the actual data you’ll use.

Here are some questions to help you think through your machine learning readiness. At Infinia ML, we use a version of this survey before helping potential clients think through the possibilities. Before you work with us – or any machine learning expert – it’s helpful to know where your team is starting from.

Machine Learning Background

What is your biggest machine learning business need? Why do you think machine learning is the solution?

Have you implemented machine learning before? If so, what frameworks did you use (e.g., Tensorflow, PyTorch, etc.)? What was the result?

How many people on your team have machine learning ability at the following levels:

  • Beginner
  • Intermediate
  • Expert

What machine learning techniques most interest you?

  • Classification
  • Numeric Prediction
  • Natural Language Processing
  • Deep Learning
  • Image Recognition
  • Other
Data Policies

What tools or programming languages does your team use to query, manipulate, and report on data?

What are the job titles on the team that works with the data? (Business Analyst, Data Scientist, Software Engineer, etc.)

Do you use third-party or public data sets? If not, are you open to the idea?

What are your data governance processes?

Your Data

Who owns the data (your company, a third party, the public domain, etc.)?

What kind of data is it?

  • Numeric
  • Text
  • Images
  • Video
  • Other (please describe):

How quickly can your team access the data?

  • Immediately
  • Upon request
  • After getting approval
  • Don’t know

How often is your data updated?

How big is the data set?

Who manages the data?

How is the data stored (local MySQL, AWS S3, Hadoop, etc.)

How sensitive is the data?

—

Are you ready to talk about advanced machine learning? So are we. We look forward to hearing from you!

The Infinia ML Interview Preparation Guide

Three people talking a whiteboard.

If you’re preparing to interview for a technical position at Infinia ML, here’s a heads up on what to expect.

We’re a machine learning company that values learning. In fact, our interview candidates sometimes say things like this:

“. . . I actually learned a lot by thinking through the problems . . .”

“This was a better process than my dissertation defense.”

 “I learned more about machine learning during this interview than I have in the last several years.”

Presentation

As part of the interview, you’ll give a 45-minute presentation to our team. The first five minutes should give an overview of your background. You’ll then have 30 minutes to cover a topic of your choosing, like a project you’ve worked on or your academic research.  In addition to technical depth, we want to understand how you think through a problem. 10 minutes of Q&A will follow your talk.

You can assume a technical audience, and supporting slides will be helpful. Feel free to bring a computer or email slides in advance to your hiring manager. 

Discussion Sessions

Beyond the presentation, we’ll have several small-group discussions to help us assess your:

  • Theoretical/conceptual understanding of machine learning
  • Ability to map theory to real-world challenges
  • Data/coding abilities
  • Cultural fit with our team

Come prepared to be challenged, but know that our team is very friendly and wants you to succeed.

Logistics

Our office is located in Research Triangle Park. The address is 1009 Slater Road, Durham, NC 27703. You can find a map on our contact page.

You can park for free in the lot next to our building. Go through the main entrance and up the elevator to the third floor. Turn left to face the hallway, turn right to go down the hallway, and you’ll find Suite 390 in an alcove on the left. Our space is WAY bigger than that door would have you believe. Please ring the bell and we’ll let you in.

We look forward to meeting you soon!

If you have any other questions, please contact your hiring manager.

Machine Learning Strategy Workshop

Infinia ML offers a one-day, in-person strategy session to help your team learn how machine learning can best add value to your business. CEO Robbie Allen leads the day along with Chief Scientist, Dr. Larry Carin, one of the world’s most prolific authors of advanced ML research.

Each Strategy Day is highly interactive and customized to add value to both business executives and technical experts. Participants will leave the Strategy Day understanding ML’s business potential – and how to distinguish hype from reality.

Participants will also receive a written assessment of their team’s most promising ML projects, including specific implementation steps.

Contact us to learn more.

Sample Agenda

9:00 – 9:15         Welcome

                                      Introducing participants’ ML experience and their goals for the day.

 

9:15 – 9:45         Machine Learning Overview

                                      Introducing ML’s development and the current state of the art.

 

9:45 – 10:15       Idea Brainstorming

                                      Building a list of potential business challenges that ML could solve.

 

10:15 – 10:30     Break

 

10:30 – 11:00     Data Overview

                                      Assessing the data required to create the identified solutions.

 

11:00 – 11:30     Data Deep Dive

                                      Analyzing the technical attributes of available company data.

 

11:30 – 12:00     Most Promising Projects

                                      Debating the highest priority projects, given impact and data.

 

12:00 – 12:15     Break

 

12:15 – 1:15        Lunch and Learn

                                      Diving deep on a chosen technical topic with Chief Scientist Larry Carin.

 

1:15 – 1:45          ML in Practice

                                      Revealing practical lessons on business deployment.

 

1:45 – 2:15          Deployment Discussion

                                      Identifying specific business pitfalls and planning to overcome them.

 

2:15 – 2:30          Break

 

2:30 – 3:00          A Glimpse of What’s Next

                                      Exploring the cutting edge of ML research with Chief Scientist Larry Carin.

 

3:00 – 3:30          Recap and Next Steps

                                      Summarizing the day’s key points, to be delivered in writing the next day.

 

3:30 – 5:00          Flex Time As Needed

 

Many companies have a five-to-ten year backlog of meaningful ML implementations that could automate complex business processes, make existing staff more efficient, and provide new capabilities that were previously not possible.

But while ML is easy to get excited about, it is hard to do well or even initiate. Contact us to learn more about hosting a Machine Learning Strategy Day.

 

Infinia ML’s First Published Papers

When Dr. Hongteng Xu joined our team, we announced that he would “continue his academic research inside Infinia ML, developing the intellectual foundation for advanced business solutions.”

Now, we’re pleased to share that Dr. Xu has already begun rolling out his research, and his work has been accepted by two prestigious conferences taking place this spring.

“Benefits from Superposed Hawkes Processes” will appear at the 21st International Conference on Artificial Intelligence and Statistics (AISTATS). The gathering is one of the top machine learning conferences, with a paper acceptance rate of just 33%.

Dr. Hongteng Xu
Hongteng Xu, Ph.D.

Dr. Xu and his co-authors, who include Infinia ML Chief Scientist Larry Carin, Ph.D., showed the potential to solve the “cold start” problem for recommendation systems. In other words, it may help the kind of systems that suggest products, movies, or music make better recommendations for new users haven’t given the system much data to work with.

Another of Xu’s papers, “Learning an Inverse Tone Mapping Network with a Generative Adversarial Regularizer,” has been accepted by the 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), a top signal processing conference.

In this paper, Xu and his collaborators proposed a new deep learning technique for inverse tone mapping, a process which helps digital images achieve better visual quality. Their innovation beats the state-of the-art tone mapping methods.

Congratulations Dr. Xu – we’re proud to have you on our team, and we look forward to sharing more of your research with the world!