Breaking News

Machine Learning in the Enterprise: Use Cases & Challenges – KDnuggets

Sponsored Post
 By Esther Rietmann, Director of Content and Programming at Data Science Salon  Enterprises leverage machine learning (ML) to improve business operations, adapt to changing business requirements, and gain insights into market trends. A 2021 ML market study indicates that 59% of all large enterprises are deploying ML solutions. ML techniques can enable fast and accurate decision-making to avoid costly corrective measures and ensure solid business reputation.
We at Data Science Salon spoke with some of the experts who will be speaking at Data Science Salon Miami Hybrid on September 21 to expand the understanding of ML in the enterprise, its key challenges and trends. Read this post to gain insightful answers that will guide you in the AI adoption journey!
The expert speakers include:

Moody Hadi, Group Manager – New Product Development & Financial Engineering at S&P Global Market Intelligence
Connie Yee, Senior Data Scientist at Bloomberg
Noelle Silver, Global Partner AI and Analytics at IBM
Sha Edathumparampil, Chief Data Officer at Baptist Health South Florida

How Do You Use Machine Learning in Your Organization?
 Several industries are embracing ML at an increasing pace to achieve business goals. Whether it’s better security, more efficient automation, or more control over finances, ML is proving to be a worthwhile investment.
According to Moody Hadi, Group Manager – New Product Development & Financial Engineering at S&P Global Market Intelligence, S&P uses ML in multiple different ways: “1) To mine unstructured data in order to create insights and signals to our clients. Typically such data sets are “alternative” data that are outside of the traditional financial data we collect. 2) To reduce the human-in-the-loop effort in order to support financial risk management workflows. 3) To automate tasks that collect data and verify their quality in order to store them and use them in 1 and 2.”
Sha Edathumparampil, Chief Data Officer at Baptist Health South Florida, explains that “ML-based solutions help improve consumer & patient experience, make operations more efficient as well as assist providers in caring for our patients.”
Connie Yee, who is a senior data scientist at Bloomberg, says that her role is to build ML solutions that help improve Bloomberg’s data and analytics that support over 300 billion requests a day: “we have a lot of opportunities to model non-traditional ML problems to use standard ML and data science algorithms. For example, the problem of ensuring the code that calls out to the services is modularized does not immediately present itself as a standard data problem. But, by adopting a graphical approach and representing code dependencies as a graph, we gained actionable insights into our application code using graph theory and embeddings.”
Noelle Silver, Global Partner AI and Analytics at IBM, sees ML as the ultimate accelerator for the enterprise. At IBM, she helps organizations perform mundane tasks like contract analysis using automation that eventually saves time for the employees and increases overall productivity for organizations.

What Are Some Main Challenges of Applying Machine Learning in the Enterprise?
 ML technologies and large-scale prediction models can be complex. At times, it becomes a challenge to translate the model outcome into explainable insights for different technical and non-technical stakeholders. 
In this regard, Connie Yee mentions that advanced ML technologies are often required to solve hard problems. However, explaining these technologies to non-technical and end-users is difficult since they are often hard to explain in terms of their design. Hence, AI interpretability and explainability is one of the main challenges. Sha Edathumparampil says on this count, “suppose you develop a great model but make it difficult for the end users to find and use it.” In that situation, “it ends up being shelf-ware.”
What is more, explainable AI is one of the key requirements for responsible AI. However, according to Noelle Silver, “creating an inclusive team to design and build the solution as well as gathering an inclusive dataset can be challenging for organizations due to the limited talent available and lack of investment in gathering inclusive data.” Therefore, clear commitment from leadership and setting a company-wide AI strategy is key to building responsible AI at scale.
While discussing the importance of the explainability aspect in the finance industry, Moody Hadi mentions the use of structural models that help analysts understand dependencies amongst the multiple independent and dependent factors. Though, he expresses dissatisfaction by saying that intuitively explaining these dependencies to end-users is challenging. For him, other industrial-scale ML challenges include hyperparameter tuning, bias-variance trade-off, and collecting high-quality, balanced training data. 

What Trends Do You See for Applying Machine Learning in the Enterprise in the Next 3-4 Years?
 Given its profound impact on various business sectors, ML is increasingly seen as an important factor in the evolution of global technology.
On the question of future trends in ML enterprise applications, Sha Edathumparampil says that no aspect of enterprises would be untouched by ML applications over the next couple of years. “AutoML frameworks, as well as ML as a service type solutions, are going to mature to the point where they are the first option for anyone looking to solve a business problem with ML solutions,” Sha adds.
Moody Hadi agrees with Sha that ML will continue to expand in usage within the enterprise, going beyond specialist use cases to wider adoption into “traditional” financial applications.
In order to fully embrace AI in the enterprise, Noelle Silver stresses the need for increasing AI literacy among executives and board members. She says, “every level of the organization will need to understand and be committed to building ML that generates positive business outcomes in the most responsible way possible.”
On the more technical side, Connie Yee mentions the emergence and wider adoption of Graph Neural Networks (GNNs) as a set of performant graph learning methods as a main trend. Apart from that, bringing together ML with knowledge graphs is another trending area in the field, in Connie’s opinion. 

Closing Remarks
 In enterprise applications, the hype around ML is snowballing. Organizations are increasingly trying to leverage ML to meet diverse business goals and positively impact different industries.  
In the conversation with leading data science experts, we learned how ML is used in the enterprise, the challenges data scientists face, and gained some insights into ML future trends. Despite many challenges, machine learning is set to disrupt major enterprises.
Join us at Data Science Salon Miami Hybrid on September 21 to hear from the experts featured in this post and 25 other thought leaders and how they apply AI and machine learning in the enterprise. Apply HERE for a complimentary conference pass (in-person or virtual) and to join technical talks, Q&A sessions with the speakers, panels discussions and network with like-minded people.