Interview with Mehrdad Mamaghani, Swedbank

Original

Anomaly detection has numerous applications in a wide variety of fields. In banking, with ever-growing heterogeneity and complexity, it is difficult to discover deviating cases using traditional investigation techniques and pre-defined scenario searches. At the Deep Learning in Finance Summit in London, this March 19 - 20, Mehrdad Mamaghani, Principal Data Scientist at Swedbank will walk attendees through how Swedbank’s deep learning models run on a state-of-the-art platform can help to detect unseen anomalies and deviations utilizing a large spectrum of features.

Mehrdad has a PhD in applied mathematical statistics from Stockholm University. Previously, Mehrdad has worked within the pharmaceutical and communication industries. At Swedbank, along with rest of the Analytics & AI group, Mehrdad and his colleagues conduct extensive work and research to better leverage the data within the bank as well as creating frameworks for more efficient and customer-oriented banking processes using deep learning techniques and advanced hardware platforms.

"At Swedbank we’re focused on everything that can optimize our day-to-day operations, and in particular our multitude of interactions with our customers. In my team, we’re some thirty curious people dedicated to a wide range of projects from discovery of next start-up unicorns to making customers’ journeys as efficient as possible. We also allocate a decent amount of time to research and deep dives into subjects that can be of potential interest for us in the future."

In the lead up to the summit, we spoke with Mehrdad to hear about his work: 

How did you begin your work in AI, and more specifically Finance?

It is difficult to draw a clear line between AI and biostatistics, which is my academic background. I guess somewhere at my previous job statistics transformed into AI: it was no longer the statistical models by themselves, there was a growing need for computer science as well as new tools and techniques. As for finance, honestly, I was attracted to the job ad. For once I thought that the ad and the job description sounded intriguing, realistic and far from generic. It turned out to be true!

 As Principal Data Scientist, what does a typical day look like for you?

We often have one or two-morning scrums and one main project that we work on. For me there’s lot of statistical coding in R and some data gym work, usually using HIVE. There’re a good number of colleagues with ideas, tips on recently published papers, blog posts or articles. These are frequently shared, discussed and sometimes implemented as pilots. Lastly, as we aim to work in a cross-disciplinary manner, we tend to spend a good chunk of our time with colleagues outside of the AI box to understand and discuss business processes and opportunities as well as disseminate knowledge on analytical techniques.

 What are the goals of your team - what problems are you trying to solve?

Our main goal is to have maximum impact on processes where mathematical and statistical analyses can enhance ways of working and minimize inefficiencies. In a sense it is about freeing up our fellow colleagues from cumbersome repetitive tasks, or guaranteeing our customers, to the greatest possible extent, a frictionless and straight-forward journey conditioned on their needs and personal preferences.

 What challenges are you currently facing in your work, and how is deep learning helping you solve them?

The subject of unbalanced data classes is something that is currently engaging us, as well as further integration of optical recognition tools into our user apps for easier handling of day-to-day errands. 

Ethics is currently a hot topic in AI,  and in finance, there's a huge amount of sensitive data. What are your thoughts on the best practices around this?

Fortunately, Swedbank has had this topic deeply understood within the bank well before GDPR became household terminology. We always try to see our work through the lens of our customers, and if we as customers – at Swedbank or elsewhere – do not find a certain use of data justifiable, we let that insight to guide us to make the right decision. However, regardless of our personal judgement, the bank has an extensive glossary on types of data and privacy aspects that we’re well-versed with. We’re working on a daily basis to clearly delineate risks based on the end purpose of an analysis and the corresponding data selection. Moreover, we have a clear and outspoken aim within our group to build democratic and unbiased models.

 The growth of AI is exponential, and countless industries are applying new models to their business to optimise efficiency and productivity as well as to solve real-world problems. What do you think the key skills are for a career in AI/DL?

Solid mathematics and statistics, specifically knowledge of building blocks and assumptions in elementary statistics such as unbiased estimation, maximum likelihood, correlation coefficients, bias/variance tradeoff etc. Programming goes without saying, here one should be open rather than over-evangelical about one specific tool or language. Being open in general to new impressions, new applications, new industries, new papers etc. is another attribute that can dramatically increase one’s understanding and open up one’s eyes to inefficiencies or problems that no one else is likely to observe. There has been a lot of good technology that has stem from cross-disciplinary dissemination.

 Which other industries are you most excited to see implement AI for a positive impact in the next 5 years?

Well, firstly, I would like to see the correct types of legislation regarding usage of AI. AI has massive potential but as any other technology it is agnostic, it can be used for good or abused. Other than that, anything and everything that pushes us to take more responsibility for the environment and our fellow citizens on the planet. To this end, the energy, transport, city planning and waste management industries could all play significant roles. One place where I want to see less AI is micro-targeted marketing, specifically when it is used in information dissemination and against minors.

Keen to learn more from Mehrdad? Register for the summit at a discounted rate now- Early Bird Discounted Passes end tomorrow, 8th Feb. 

Original

AI Deep Learning in Finance Summit FinTech Financial Compliance Retail Finance Finance Financial Forecasting


0 Comments

    As Featured In

    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original

    Partners & Attendees

    Intel.001
    Nvidia.001
    Acc1.001
    Rbc research.001
    Mit tech review.001
    Graphcoreai.001
    Twentybn.001
    Forbes.001
    Kd nuggets.001
    Ibm.001
    Facebook.001
    Maluuba 2017.001
    This website uses cookies to ensure you get the best experience. Learn more