Software Engineering Radio - the podcast for professional software developers avatar

Dexa/Software Engineering Radio - the podcast for professional software developers

Learn more
Published Dec 14, 2023

SE Radio 594: Sean Moriarity on Deep Learning with Elixir and Axon

Explore the transformative potential of deep learning with Sean Moriarity as he delves into designing fraud detection models using Axon, an innovative framework enhancing Elixir's capabilities, highlighting profound advantages over traditional methods and addressing challenges in feature engineering and model evaluation.
Episode Highlights
Software Engineering Radio - the podcast for professional software developers logo

Topics covered

Popular Clips

Episode Highlights

  • Deep Learning

    Deep learning excels in handling high-dimensional data, making it ideal for complex problems like image recognition and natural language processing. explains that deep learning simplifies feature engineering by transforming raw data into structured representations, unlike traditional methods that require manual feature extraction 1. He emphasizes that deep learning is best suited for problems where accuracy and precision are paramount, while simpler tasks might benefit from straightforward heuristics 2.

    Deep learning comes in when you're dealing with what's essentially called the curse of dimensionality.

    ---

    This approach allows for the automatic learning of features, reducing the need for extensive pre-processing 3.

       

    Feature Engineering

    Feature engineering is crucial in machine learning, yet deep learning automates much of this process, enhancing efficiency. notes that deep learning models can automatically discern relevant features from raw data, such as images, without extensive manual input 1. In fraud detection, for instance, deep learning can identify patterns and anomalies in transaction data, even when dealing with unbalanced datasets 4.

    One of the strengths is that I don't necessarily need to make a ton of choices about what inputs I pass into my deep learning model.

    ---

    This capability is particularly valuable in scenarios where traditional feature engineering would be labor-intensive and less effective 5.

       

    Machine Learning

    Traditional machine learning relies on explicit rules and feature extraction, while deep learning offers a probabilistic approach to complex problems. highlights that deep learning is essential when traditional methods fall short, especially in high-dimensional spaces 6. He contrasts this with simpler tasks where basic heuristics might suffice, emphasizing the importance of choosing the right tool for the job 2.

    Simple heuristics might do much better.

    ---

    This distinction underscores the necessity of machine learning in modern applications, where adaptability and precision are crucial 7.

Related Episodes