For data scientists and engineers, it is fairly common to reuse features learned by pre-trained models (transfer learning). When this is done in practice, several questions arise - what model architecture do I use? Do I use the entire model or a subset of the model? What patterns do layers in each model learn and what tasks are they best suited to? Building intuition on how to navigate these choices requires extensive experimentation which can be compute and time intensive.
To help address this, ConvNet Playground provides an environment to explore (the results of) combinations of these choices in a…
Handtrack.js library allows you track a user’s hand (bounding box) from an image in any orientation, in 3 lines of code.
In August 2018, I had the opportunity to attend the 2018 Deep Learning Indaba where Google generously provided access to TPUs (v2) to all participants! I finally came around to spending some time getting familiar with running ML experiments on TPUs and started out with training a GAN to generate images based on a custom dataset I have been curating — the African masks dataset. TPUs provide some serious compute horsepower (at par with high end GPUs) that enable fast turnaround for training experiments . Having access to a TPU allowed me explore some questions related to training GANs —…
Perhaps the best deep learning meeting I have attended. Heres why.
Some months ago, I decided to build more experience in core AI and explore its applications in my research domain (HCI). In addition to taking online courses, personal experimentation, sharing expertise, an important part of this journey has been to engage with the community.
And so I was really excited at the opportunity to attend the 2018 deep learning Indaba (Indaba means “meeting” in Zulu)! My main goals were to learn from the rather high quality speaker panel (Omoju Miller, Naila Murray, Katja Hofmann, David Silver, Kyunghyun Cho, Jeff…
Increased enterprise adoption, new tools/extensions, new ways to improve datascience.
At the just concluded JupyterCon 2018 conference, it really was all about leveraging the power of Project Jupyter for collaborative, extensible, scalable, and reproducible data science. I got to attend tutorials, talks, keynotes and a poster session (where I presented updates on my work on automated visualization) across 3 days. A TLDR summary of some highlights I found interesting are below:
Control the game paddle by waving your hand in front of your web cam.
With the TensorFlow object detection api, we have seen examples where models are trained to detect custom objects in images (e.g. detecting hands, toys, racoons, mac n cheese). Naturally, an interesting next step is to explore how these models can be deployed in real world use cases — for example, interaction design.
In this post, I cover a basic body-as-input interaction example where real time results from a hand tracking model (web cam stream as input) is mapped to the controls of a web-based game (Skyfall)…
The conference theme was “AI and Deep Learning”.
4 days of excellent talks, keynote, demos, posters and interactions at the NVIDIA GPU Technology Conference. As a HCI researcher interested in applied AI, it was exciting to learn about advances in AI hardware (GPUs), technical talks covering both the science of AI and industry use cases of AI. This post summarizes my notes from the conference keynote and technical sessions I attended.
TLDR — some highlights
We formulate data visualization as a sequence to sequence translation problem.
TLDR; We train a model that can take in a dataset as input and generates a set of plausible visualization as output.
Summary of talks from AAAI18 that cover topics in Computer Vision, Machine Learning/Deep Learning(Catastrophic Forgetting), Learning Representations, Knowledge Graphs and Applied AI in general.
This post document contains notes on sessions I attended at the just concluded Artificial Intelligence Conference (AAAI 2018, New Orleans Louisiana). The selection of talks where based on my interests from a HCI and applied AI perspective. These include Human aspects of AI, Vision, Machine Learning/Deep Learning(Catastrophic Forgetting), Learning Representations, Knowledge Graphs and Applied AI in general. I welcome feedback, corrections (typos!) and discussions - please get in touch (@vykthur). For those interested in an additional…
TLDR: We train a model to detect hands in real-time (21fps) using the Tensorflow Object Detection API.