Better Interpretability Leads to Better Adoption Is your highly-trained model easy to understand? A sophisticated machine learning algorithm usually can produce accurate predictions, but its notorious “black box” nature does not help adoption at all. Think about this: If you ask me to swallow a black pill without telling me…

Image classification is the task to recognize an image. It is also called image recognition. Computer scientists have been innovative in extracting meaning from images. Its history is fascinating, though most people don’t know much about it. For this reason, I am going to tell you the stories of innovation…

Many machine learning algorithms are complicated and not easy to understand, even though they have rendered an impressive level of accuracy. As humans, we must be able to fully understand how decisions are being made so that we can trust the decisions of AI systems. We need ML models to…

You receive a data science interview opportunity from your dream company. You have surveyed many the-top-50-question types of articles but still feel uncertain. Since there are already many similar articles, why do I dare to add an article to this crowded topic? In this article, I re-write many ordinary answers…

In supervised learning, we use R-squared, ROC, Precision-call, or F-sore to evaluate performance during model training. How is a Large Language Model evaluated? Large Language Models are Transformer-based models built on complex neural networks and fundamentally follow supervised learning framework. They still apply the typical train-test-validation data split. …

Why can Large Language Models (LLMs) answer questions, do book reports, draft notes, or summarize a document? An important reason is the data that they were trained on, or fine-tuned with. This post helps you to understand those widely used datasets that are known in the LLM community. While you…

Machine learning uses a lot of discrete distributions such as the Bernoulli, Binomial, and Multinomial distributions to solve problems. Two related discrete distributions, the Beta and Dirichlet distributions are less known but are widely used in data science. The Dirichlet distribution is especially important in Natural Language Processing (NLP). The…

In Chapter 1, we talked about supervised learning can target better for known outliers, and unsupervised learning can explore new types of outliers. Can we take the advantage of both supervised and unsupervised learning? Specifically, since supervised learning achieves better precision in general, and the outlier scores from unsupervised learning…

Classification problems are often solved using supervised learning algorithms such as Random Forest, Support Vector Machine, Logistic Regressor, and so on. Supervised learning algorithms require a known target to build a model. However, it is often the case that we only see normal data patterns but not rare events. The…