Ari Seff will present his Pre FPO "Structured Generative Modeling for Engineering and Design" on Friday, January 15, 2021 at 2pm via zoom. Link: [ https://princeton.zoom.us/j/95701063699?pwd=WE96SmpsZXlwUUtYUzE5THFqeWNGdz09 | https://princeton.zoom.us/j/95701063699?pwd=WE96SmpsZXlwUUtYUzE5THFqeWNGdz09 ] Passcode: ari2021 The members of his committee are as follows: Readers: Ryan Adams, Szymon Rusinkiewicz Examiners: Tom Griffiths, Olga Russakovsky, Abigail Doyle (Chemistry) Talk abstract follows below. All are welcome to join. Abstract: Generative modeling is poised to make a large impact on the fields of engineering, design, and scientific discovery. Integral to these fields are discrete domains, such as molecules and graphs. In contrast to the Euclidean spaces where generative modeling has recently thrived (e.g., images and audio), discrete domains often exhibit irregular structure, strict validity constraints, and non-canonical representations, presenting unique challenges to modeling. In addition, machine learning as applied to engineering still lacks maturity, and as a result there is a need for curated datasets, benchmarks, and shared pipelines. First, we discuss a novel generative model for discrete domains, known as reversible inductive construction, that addresses some of the above challenges. Building off of generative interpretations of denoising autoencoders, the model employs a Markov chain where transitions are restricted to a set of local operations that preserve validity. This approach constrains the generative model to only produce valid objects, requires the learner to only discover local modifications to the objects, and avoids marginalization over an unknown and potentially large space of construction histories. Next, we present SketchGraphs, a dataset and processing pipeline that contains 15 million real-world CAD sketches paired with their ground truth geometric constraint graphs. Unlike previously available CAD data, explicit supervision is provided regarding the designer-imposed geometric relationships between primitives, making the typically latent construction operations directly accessible. We demonstrate and establish benchmarks for two use cases of the dataset: generative modeling of sketches and conditional generation of likely constraints given unconstrained geometry. Lastly, we propose a method for CAD sketch inference from images, leveraging a conditional generative model that consumes a noisy observation (e.g., hand-drawn) of a design and samples a plausible geometric constraint graph. Evaluation of this model on SketchGraphs indicates its potential to enable new design workflows.