← THE FIELD RECORD
SECOND RUN AT UT DALLAS

Five days of deep learning at UT Dallas, then a Saturday for high-school students

7-12 January 2019 · The University of Texas at Dallas, Richardson, Texas · IMPACT

Eight months after the first visit, the lab returned to the Department of Computer Science for a five-day bootcamp, then stayed an extra day to teach high-school students.

The full cohort standing together across the front of the lecture room on the last day, benches and open laptops in the foreground, whiteboards and projection screens behind them.

The bootcamp ran from Monday 7 January to Friday 11 January 2019 on the UT Dallas campus in Richardson, Texas. Tarry Singh drew the whole week on a blackboard on 4 January, three days before it started. Five columns, one per day, a theory block at the top of each and a coding block underneath it. The header reads: Jan 7-12 2019, UT Dallas, deep dive in deep learning.

Day one was the fundamentals, Numpy and Pandas, and a first look at computer vision. Day two was convolutional networks, MNIST and a CIFAR exercise. Day three went to advanced CNN theory and a skin cancer image classifier built on InceptionV3. Day four was recurrent networks, with time series exercises on Bitcoin prices and household power consumption. Day five ran EEG classification with an RNN, and the notebook folders still hold the data files each of those days used.

Photographs from the middle of the week show the hall full. Long benches, a laptop open at almost every seat, and live code on the screen at the front more often than slides. Teaching moved between the podium and the whiteboard, where a dense layer with tanh activations was drawn out by hand next to three input matrices and the words your first batch.

The May 2018 visit had been a lecture series. This one was built around a project a day, and it carried two things the first one did not. The first was an open AI Playbook session for data science professionals and executives, walking an idea through four gates: assessment, development, test, launch. The second was Saturday 12 January, given over to high-school students. A talk from ten to twelve, coding a first neural network from one to three, then an open session at three where they planned a project of their own.

Certificates came from the UT Dallas Department of Computer Science. They were handed over one at a time, in front of that same whiteboard, with the diagram from the last lecture still on it.

FROM THE FIELD · 08 FRAMES
Tarry Singh writing at the whiteboard with his back to the room, drawing a dense layer with tanh units next to three input matrices, students watching from the benches.
Tarry Singh writing at the whiteboard with his back to the room, drawing a dense layer with tanh units next to three input matrices, students watching from the benches.
Wide view from the back of the hall: two instructors at the front podium, Python code projected on the screen above them, students at laptops along every bench.
Wide view from the back of the hall: two instructors at the front podium, Python code projected on the screen above them, students at laptops along every bench.
Rows of students at the tiered benches with laptops open, looking towards the front of the hall.
Rows of students at the tiered benches with laptops open, looking towards the front of the hall.
Certificate handover in front of the whiteboard: a participant holds a UT Dallas Department of Computer Science certificate beside Tarry Singh, the lecture diagram and the words your first batch still on the board.
Certificate handover in front of the whiteboard: a participant holds a UT Dallas Department of Computer Science certificate beside Tarry Singh, the lecture diagram and the words your first batch still on the board.
A slide headed Vector to Sequence on the front screen of the hall, the instructor at the podium below it and students in the foreground.
A slide headed Vector to Sequence on the front screen of the hall, the instructor at the podium below it and students in the foreground.
Tarry Singh speaking at the front of the room, a matrix written on the whiteboard behind him, heads of the audience in the foreground.
Tarry Singh speaking at the front of the room, a matrix written on the whiteboard behind him, heads of the audience in the foreground.
The handwritten plan for the week on a blackboard, headed Jan 7-12 2019, UT Dallas, deep dive in deep learning, with five day columns from coding fundamentals through to advanced RNN.
The handwritten plan for the week on a blackboard, headed Jan 7-12 2019, UT Dallas, deep dive in deep learning, with five day columns from coding fundamentals through to advanced RNN.
WITH
The University of Texas at Dallas, Department of Computer Science
← EARLIERBusiness Executive to AI LeaderLATER →What an AI company actually does, explained in Assen