Generative neural networks can project the variability in the data onto a manifold and enables exploration of the design space by means of interpolation and extrapolation in this latent space. While there is a vast literature on non-sequential data, this problem hasn’t received much attention for time-series data. In this project, we aim to develop a sequence model that is able to infer interpretable features from sequences. Particularly we will be using digital ink data, i.e., handwriting and hand-drawn sketches. It is a type of data modulated by the personalized effects of fine-grained human motion. Such a model is expected to infer the personalized features such as slanting of handwriting, size of the letters or presence of ligatures, and to allow controlling appearance of generated handwriting samples by means of changing these features.
The ideal candidate for this position should have some background in deep learning. An existing code base in Tensorflow to train and evaluate models will be provided.
Generative neural networks can project the variability in the data onto a manifold and enables exploration of the design space by means of interpolation and extrapolation in this latent space. While there is a vast literature on non-sequential data, this problem hasn’t received much attention for time-series data. In this project, we aim to develop a sequence model that is able to infer interpretable features from sequences. Particularly we will be using digital ink data, i.e., handwriting and hand-drawn sketches. It is a type of data modulated by the personalized effects of fine-grained human motion. Such a model is expected to infer the personalized features such as slanting of handwriting, size of the letters or presence of ligatures, and to allow controlling appearance of generated handwriting samples by means of changing these features. The ideal candidate for this position should have some background in deep learning. An existing code base in Tensorflow to train and evaluate models will be provided.
Each year the IDEA League offers the students of its partner universities over 180 monthly grants for a short-term research exchange. In general, these grants are awarded based on academic merit. For more information visit http://idealeague.org/student-grant/