HomeMachine LearningMachine Learning NewsReducing the expectation-reality gap in Machine learning

Reducing the expectation-reality gap in Machine learning

In every industry today, machine learning (ML) is essential. To promote creativity and long-term growth, business executives are pushing their technical teams to expedite the adoption of machine learning throughout the company. However, there is a gap between what business executives truly expect from large-scale machine learning deployment and what engineers and data scientists can really create quickly and efficiently.

The majority of business executives in a Forrester study, which was commissioned by Capital One and released today, expressed excitement about implementing machine learning (ML) across the organization. However, data scientist team members claimed they lacked some essential tools to develop ML solutions at scale. Companies would be thrilled to take advantage of machine learning (ML) as a plug-and-play opportunity, whereby they could “just input data into a black box and valuable learnings emerge.” Engineers that manage corporate data to create machine learning models are aware that it goes far beyond that. There are compliance, regulatory, and security requirements to meet, and the data may be unstructured or of low quality.

Establishing open communication between teams is the first step towards narrowing the expectation-reality gap, but there is no quick fix. Business executives can then start implementing ML throughout the company. Democratization entails providing strong machine learning (ML) tools and ongoing training and education to both technical and non-technical teams. While data scientists have access to the powerful development platforms and cloud infrastructure they need to effectively create machine learning applications, non-technical teams can make better business decisions with the help of easy-to-use data visualization tools. These democratization techniques have helped Capital One, a company with over 50,000 employees, scale machine learning.

There’s less of a divide between the technical and business teams when everyone is motivated to use machine learning to make the company successful. Thus, how can businesses start democratizing machine learning? These are some best practices for enabling everyone in the company to benefit from machine learning.

Enable your creators

Today’s top engineers are not just technical wizards; they are also critical collaborators with product designers and specialists, as well as creative thinkers. Companies should give opportunities for tech, product, and design to collaborate on common objectives in order to promote greater collaboration. Because ML use can be siloed, putting an emphasis on collaboration can be a critical cultural element of success, according to the Forrester study. Additionally, it will guarantee that products are developed from a technical, human, and business standpoint.

Leaders should also find out what resources engineers and data scientists require in order to successfully expedite the delivery of machine learning solutions to the company. Forrester reports that 67% of participants concur that the cross-enterprise adoption of machine learning is being hampered by a lack of user-friendly tools. An underlying technological infrastructure supporting ML engineering should be compatible with these tools. Avoid creating a “hurry up and wait” environment for your developers, where they can develop an ML model in the sandbox staging area, but can’t deploy it right away because they lack the infrastructure and compute to put the model into production. For ML training environments to function, a strong cloud-native multitenant infrastructure is essential.

Empower your employees

Any company can become a data-driven organization by empowering every employee, regardless of role, with the ability to leverage machine learning (ML). Companies can begin by allowing controlled access to data for their staff members. Provide teams with low-code/no-code tools to analyze data and make business decisions after that. Obviously, the goal of human-centered design should be considered when creating these tools to ensure ease of use. In an ideal world, a business analyst could apply machine learning (ML) functionality via a clickable interface, upload a data set, and produce outputs that could be used right away.

A large number of workers are keen to learn more about technology. Leaders should give teams in the organization a variety of opportunities to acquire new competencies. Many technical upskilling initiatives have proven successful for Capital One, including our Tech College, which provides instruction in seven technology-related disciplines in line with our business imperatives; our Machine Learning Engineering Programme, which equips recent college graduates with the skills needed to launch a career in machine learning and artificial intelligence; and the Capital One Developer Academy, which trains non-computer science majors for careers in software engineering. Sixty-four percent of those surveyed by Forrester agreed that a lack of training was impeding the use of machine learning in their companies. Fortunately, upskilling is something that every business can provide by encouraging seasoned employees to guide younger employees.

Measure and celebrate success

Encouraging data-driven decision-making across the entire organization can be achieved by democratizing machine learning. It is imperative to assess the effectiveness of democratization efforts and consistently enhance areas that require improvement. In order to measure the effectiveness of machine learning democratization, executives can examine which data-driven choices made via the platforms produced quantifiable business outcomes, like more income or new clients. We have quantified the savings that customers have experienced at Capital One, for instance, thanks to our machine learning innovations in anomaly and change point detection that have enabled card fraud defense.

Collaboration, measured accountability, and teamwork are the cornerstones of any successful machine learning democratization programme. Technical teams can receive feedback on what features would make their jobs easier from business users of machine learning (ML) tools. Technical teams can ask for tools and training to help them succeed, as well as discuss the difficulties they encounter when developing new iterations of their products.

When end users ultimately gain from a single, human-centered vision for machine learning (ML), business executives and technical teams come together. Better products and services that satisfy customers can be produced by a business using data-driven learnings. To create a forward-thinking company that innovates with potent data insights, implement a few best practices to democratize machine learning throughout the organization.

Source link


Most Popular