Our first Spanish webinar featured the highly-accomplished Héctor Cuesta, who is the Director of Product Management at Kueski, an adept Data Scientist, and the founder of Dataxios, an AI research company. During the webinar, Héctor delved into the significance of machine learning in the banking industry, and he is also recognised for his work in authoring the book, Practical Data Analysis.
What is the significance of machine learning?
To initiate the discussion, Héctor posed a thought-provoking question.
When we aim to utilize an algorithm for a purpose beyond its original design, how can we ensure that it can adapt to a variety of scenarios?
Generally, algorithms have limited capabilities in handling varying inputs. Nonetheless, in unanticipated circumstances, machine learning can be leveraged to generalize the situation, allowing for proper management and anticipation of any unanticipated developments. But how does this method operate?
Three significant breakthroughs in machine learning’s evolution
Héctor has pinpointed three crucial episodes in the development of machine learning, comprehending which can offer valuable knowledge about how this technology operates.
- In an academic paper published in 1949, Claude Shannon introduced the Minimax algorithm, which is utilized by computer programs that play chess to anticipate their opponents’ moves. This algorithm utilises the fixed rules of chess not only to generate moves based on its knowledge but also to forecast the actions of its adversary.
- In 2011, IBM Watson’s triumph in Jeopardy! brought about a significant transformation in machine learning, resulting in its current state, where it can forecast outcomes based on predetermined criteria in both structured and unstructured environments. This capability was showcased again in 2022, when the same algorithm adapted to the latest environment and could make predictions founded on its prior performance in the game show Jeopardy!.
- Google’s AlphaGo software made history in 2022 when it defeated the Go grandmaster Lee Sedol in an unpredictable match. This accomplishment was extraordinary due not only to the game’s complexity but also because the technology employed was easily accessible, relatively low-cost, and straightforward to implement. This technology’s potency is such that it can learn autonomously. Its implications are vast, inaugurating a new era of machine learning.
Comprehending machine learning need not be a challenging task. Fundamentally, it can be distilled into three words: observation, learning, and prediction. However, as one delves into the more intricate facets of this discipline, such as its potential applications, the subject becomes increasingly complex and intriguing.
What role does machine learning play?
Over the past decade, machine learning has advanced significantly, giving rise to several subcategories that cater to its diverse applications. Héctor has identified the three primary domains of machine learning:
Computer Programs Requiring Human Supervision: These algorithms may be trained using a database, enabling them to anticipate specific outcomes, similar to how fraud detection algorithms operate as demonstrated in this article.
Unsupervised Algorithms: Unlike supervised algorithms that depend on predefined values, unsupervised algorithms employ a comparative approach to detect commensurability traits in previously unexplored settings, utilising the obtainable data. This method is frequently referred to as ‘clusterisation’.
Reinforcement Learning: Drawing from previous iterations, reinforcement learning offers the unique benefit of incremental learning.
The Confluence of Artificial Intelligence and Financial Technology
One of the most attractive attributes of machine learning is its precision, which enables financial experts to meet their niche market’s demands.
Héctor is intrigued by the specific methods through which machine learning can enhance fintech.
By enabling the Financial Technology sector to pose specific questions, machine learning serves as a potent tool for identifying, recruiting, and selecting the most appropriate candidates.
Machine learning can provide answers to questions such as:
Which category of customers does this particular one belong to?
What is the relevance of this characteristic in this particular context?
How does this variation resemble the original?
What will be the long-term value of specific attributes?
What is the desired outcome, and what is the rationale behind it?
Incorporating such enquiries into a financial plan would allow for the development of a customised strategy to fulfil customers’ requirements. For instance, if a bank was aware of a customer’s demographic, it could enhance its ability to foresee their behaviour. Thus, a salesperson could meet with someone classified as a “senior employer” or “financially secure” to introduce them to retirement savings or similar plans.
Questioning “what characteristic resembles the original” can assist in directing the algorithm to search for, categorise, and propose benefits that are comparable or compatible with what the customers initially received. As an illustration, after approving a mortgage, a machine learning algorithm may suggest home insurance to a consumer.
Eventually, this data is utilised to devise services and products for fintech. These encompass tools that assist in financial decision-making for both customers and financial institutions; fraud and security tools; chatbots customised to each user’s financial interactions; and credit-evaluation platforms that can determine the suitability of a customer for a loan with just one click. These are just a few instances.
In the fast-evolving landscape of fintech, machine learning models can provide the financial sector with precise and thorough analysis and forecasting of data.
Are you a software developer looking to advance your career? Share your hiring request with us at https://www.works.so/hire-talent!
