
John Thickstun
Contact: thickstn at cs.washington.edu I am a PhD candidate in Computer Science & Engineering at the University of Washington, coadvised by Sham Kakade and Zaid Harchaoui. I completed my undergraduate degree in Applied Mathematics at Brown University, where I was advised by Eugene Charniak and Björn Sandstede. My current research interests include generative models, sampling, time series, and applications to music. My research has been supported by a 2017 NSF graduate fellowship, and a 2020 Qualcomm innovation fellowship. I will be completing my PhD in Summer 2021, and joining Stanford in Autumn as a postdoc working with Percy Liang. My CV is available here. 
Teaching Autumn 2020: CSE 599 Generative Models Research Directions  
Source Separation and Conditional Sampling 
We can use a generative model as a prior for decomposing a linear mixture of sources into its constituent parts. We demonstrate this on the left for linearly superimposed images of churches and bedrooms. The idea is to train train two generative models: one that estimates the likelihood of the distribution over images of churches, and another that does the same for images of bedrooms. We can separate a mixture of images by decomposing it into two images that are likely under the priors for churches and bedrooms respectively, that satisfy the constraint that decomposition must sum to the given mixture. This can be interpreted as conditional sampling from the posterior distribution over sources given a mixture. Our work on visual source separation has appeared in ICML 2020. Extensions of these results to audio separation and more general conditional sampling problems are in preparation. [Conference Paper] [Recorded Presentation] [Code] 
Generative Modeling of Musical Scores 
Generative models are powerful tools for revealing structure in data. Features learned by fitting an unsupervised generative modeling objective can be transferred to other tasks. Or, as see in the source separation project above, we can directly leverage these generative models as priors. A fun aspect of these models is that you can sample from them; a generative model over musical scores can be turned into a kind of automatic music composer (see left). Musical scores are highly structured, heterogenous objects. Their twodimensional structure is reminiscent of visual data, but their timeseries structure and sparsity are reminiscent of language. In contrast to both language and visual imagery, the number of scores in a particular musical genre is limited. This makes score modeling an inherently lowresource learning problem. In work appearing in ISMIR 2019, we discuss domainspecific modeling choices that help maximize what we can learn from limited data. [Conference Paper] [Demos] [Code] 
MusicNet and Music Transcription

Musical scores comprise a dense set of labels on performances of classical western music. These labelings are analogous to semantic segmentations of visual imagery. However, in order to use a score as a segmentation map, we must first align it to particular audio recording by warping precise timings in the score onto expressive timings chosen by the performers. This is the music alignment problem. Using an alignment algorithm, we created the MusicNet dataset by aligning scores to a collection of freelylicensed recordings. We can use these labels to, for example, train an automatic music transcription system. In work appearing at ICASSP 2018, we describe a stateoftheart transcription model trained using MusicNet that todate (2019) is the bestperforming algorithm in the MIREX transcription challenge. [Conference Paper] [Code] 
Dissertation Leveraging Generative Models for Music and Signal Processing. Publications and Preprints MAUVE: HumanMachine Divergence Curves for Evaluating OpenEnded Text Generation. Rethinking Evaluation Methodology for AudiotoScore Alignment. Parallel and Flexible Sampling from Autoregressive Models via Langevin Dynamics. Faster Policy Learning with ContinuousTime Gradients. An Information Bottleneck Approach for
Controlling Conciseness in Rationale Extraction. Source Separation with Deep Generative Priors. Convolutional Composer Classification. Coupled Recurrent Models for Polyphonic Music Composition. Invariances and Data Augmentation for Supervised Music Transcription. Learning Features of Music from Scratch (MusicNet). Notes and Tutorials A brief mathematical introduction to GAN's The Transformer model in equations Information theory: an alternative introduction with applications to concentration of measure Conditional Random Fields as a generalization of logistic regression Some notes on HilbertSchmidt operators Estimating the Shannon capacity of a graph Thoughts on proof assistants with companion code Finance Three perspectives on the BlackScholes formula:
Heuristics for manipulating stochastic differential equations Negative probabilities in the binomial option pricing model Probability The fast JohnsonLindenstrauss transform Probability densities from a measuretheoretic perspective Fun stuff 