I describe the data science life cycle, a methodology for effectively developing and deploying data-driven projects.
In this fifth post in a series on how to build a complete machine learning product from scratch, I describe how to deploy our model and set up a continuous integration system.
In this fourth post in a series on how to build a complete machine learning product from scratch, I describe how to error analyze our first model and work toward building a V2 model.
In this third post in a series on how to build a complete machine learning product from scratch, I describe how to build an initial model with an associated training/evaluation pipeline and functionality tests.
In this second post in a series on how to build a complete machine learning product from scratch, I describe how to acquire your dataset and perform initial exploratory data analysis.
In this first post in a series on how to build a complete machine learning product from scratch, I describe how to setup your project and tooling.
After analyzing 1000+ Y-Combinator Companies, I discover there's a huge market need for more engineering-focused data practitioner roles.
Deploying a State-of-the-Art Question Answering System With 60 Lines of Python Using HuggingFace and Streamlit
I show how you can deploy a performant question-answering system with just 60 lines of Python.
I describe in-depth the courses necessary for a 4-year undergraduate degree in artificial intelligence, assuming you step onto campus tomorrow.
Following an eventful week at NeurIPS 2019, I summarize some key trends in machine learning today.
In which I describe how I built a deep learning computer starting from nothing but a pile of hardware components and a dream.
I provide an aggregated and comprehensive list of tutorials I have made for fundamental concepts in machine learning.
I discuss decision trees which are very powerful, general-purpose models that are also interpretable.
I discuss market basket analysis, an unsupervised learning technique for understanding and quantifying the relationships between sets of items.
I discuss the k-nearest neighbors algorithm, a remarkably simple but effective machine learning model.
In which I describe commonly used techniques for evaluating machine learning models.
In which I discuss the technique of ensembling which is used to improve the performance of a single machine learning model by combining the power of several other models.
I discuss strategies such as cross-validation which are used for selecting best-performing machine learning models.
In which I discuss regularization, a strategy for controlling a model's generalizability to new datasets.
In which I give a primer on principal components analysis, a commonly used technique for dimensionality reduction in machine learning.
In which I discuss feature selection which is used in machine learning to help improve model generalization, reduce feature dimensionality, and do other useful things.
In which I describe the bias-variance tradeoff, one of the most important concepts underlying all of machine learning theory.
Following an eventful week at ACL 2019, I summarize key trends and takeaways to look for in NLP.
In which we investigate K-means clustering, a common unsupervised clustering technique for analyzing data.
I describe recurrent neural networks, a deep learning technique geared for tasks such as natural language processing.
We discuss convolutional neural networks, a deep learning model inspired by the human visual system that has rocked the state-of-the-art in computer vision tasks.
We investigate a motley of interesting properties and goodies of neural networks.
We begin a deep dive into deep learning by investigating feedforward neural networks.
We discuss support vector machines, a very powerful and versatile machine learning model.
I describe Naive Bayes, a commonly-used generative model for a variety of classification tasks.
I describe logistic-regression, one of the cornerstone algorithms of the modern-day machine learning toolkit.
I describe the basics of linear regression, one of the most common and widely used machine learning techniques.
We discuss the Transformer, a purely attention-based architecture that is more performant, more efficient, and more parallelizable than recurrent network-based models.
I describe ELMo, a recently released set of neural word representations that are pushing the state-of-the-art in natural language processing pretraining methodology.
I dive into the details of Fast R-CNN, an extension to the original R-CNN model that boasted 9x speedup over its predecessor as well as state-of-the-art object detection results.
A discussion of R-CNN, a historic object detection architecture combining region proposals with convolutional neural networks.
A discussion of fundamental deep learning algorithms people new to the field should learn along with a recommended course of study.
Given all the recent buzz around artificial intelligence, I discuss three reasons for why we are seeing such widespread interest in the field today.
A discussion of the most important skills necessary for being an effective machine learning engineer or data scientist.
In light of a particularly exhausting software job interview cycle, I discuss the general format of such interviews and propose a playful challenge for any people going through a similar experience.
Following my attendance at the 18th Annual Meeting on Discourse and Dialogue, I summarize the most promising directions for future dialogue research, as gleaned from discussions with other researchers at the conference.
To help spur conversational assistant research, we release a corpus of 3,031 grounded, multi-turn dialogues in three distinct domains appropriate for an in-car assistant: calendar scheduling, weather information retrieval, and point-of-interest navigation.