Honors & Awards
Walter V. and Idun Berry Postdoctoral Fellow, Stanford University (08/2015)
Bachelor of Arts, Brown University (2008)
Bachelor of Science, Brown University (2008)
Doctor of Philosophy, Stanford University, MATH-PHD (2013)
Bachelor of Science, Brown University, BS Math, BA Econ(5 yr program) (2008)
Nicholas Haber, Catalin Voss. "United States Patent Application 14/275851 Systems and methods for detection of behavior correlated with outside distractions in examinations"
Nicholas Haber, Catalin Voss. "United States Patent Application 61/821,921 System and Method for Analysis of Visual Viewer Reactions to Video Content. US Application"
Current Research and Scholarly Interests
My research lies at the intersection of medicine, artificial intelligence, and mathematics. Most of my current activities are devoted towards a collaborative project with a multidisciplinary group of researchers, aimed at developing a wearable device with automatic facial expression recognition technology for the purpose of autism therapy. Many on the autism spectrum struggle in reading facial expressions, and the standard cognitive behavioral therapy for this essentially amounts to flashcards – examples of facial expressions for memorization, without larger context. This therapy works, often, but it is a slow, painstaking process. In the creation of such a device, we look to bring this learning effort to the real world, allowing the user to practice recognizing facial expressions of their family and friends with the help of cues and hints from the software. One hypothesis is that a system which simply informs the user that the person they are talking to looks happy, surprised, or sad will lead to much more rapid development, but it could also be the case that more nuanced help, such as being able to tell when the other person is engaged or confused or nervous, will produce the most powerful learning effects. It is difficult to predict what will happen when such therapeutic tools are deployed in the home, and we are very excited to see the sort of data we will observe in upcoming studies.
My particular contributions to this project primarily involve the core expression recognition. I design and use algorithms that learn how to recognize facial expressions from video and image data. So-called affective computing is a growing field of study with many difficulties. The art of teaching a computer to recognize the facial expressions of a person it has never seen before is very imperfect, and in a project such as this, it is imperative that recognition succeeds nearly all of the time. I thus draw on my background in mathematics and machine learning to explore new methods by which we might create more accurate recognition. Towards this, I have been working on convolutional neural network methods, and I am interested in creating novel related architectures and in exploring the properties of convnet training.
More broadly, I see myself as a mathematician looking to bring his skills over to medicine in order to make impactful contributions to diagnosis and therapy. For instance, I have been advising an effort by researchers to develop machine learning classifiers that discern those on the autism spectrum from those with ADHD using phenotypic data. This could potentially lead to more rapid, cheaper diagnoses.
I maintain an active interest in mathematics, both in the sorts of research I have pursued throughout my career (mathematical physics, in particularly that which pertains to the foundations of quantum theory) and in the general promotion of mathematical literacy in the sciences.
Crowdsourced validation of a machine-learning classification system for autism and ADHD.
2017; 7 (5)
Autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) together affect >10% of the children in the United States, but considerable behavioral overlaps between the two disorders can often complicate differential diagnosis. Currently, there is no screening test designed to differentiate between the two disorders, and with waiting times from initial suspicion to diagnosis upwards of a year, methods to quickly and accurately assess risk for these and other developmental disorders are desperately needed. In a previous study, we found that four machine-learning algorithms were able to accurately (area under the curve (AUC)>0.96) distinguish ASD from ADHD using only a small subset of items from the Social Responsiveness Scale (SRS). Here, we expand upon our prior work by including a novel crowdsourced data set of responses to our predefined top 15 SRS-derived questions from parents of children with ASD (n=248) or ADHD (n=174) to improve our model's capability to generalize to new, 'real-world' data. By mixing these novel survey data with our initial archival sample (n=3417) and performing repeated cross-validation with subsampling, we created a classification algorithm that performs with AUC=0.89±0.01 using only 15 questions.
View details for DOI 10.1038/tp.2017.86
View details for PubMedID 28509905
- The Feynman Propagator on Perturbations of Minkowski Space COMMUNICATIONS IN MATHEMATICAL PHYSICS 2016; 342 (1): 333-384
Use of machine learning for behavioral distinction of autism and ADHD.
Although autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) continue to rise in prevalence, together affecting >10% of today's pediatric population, the methods of diagnosis remain subjective, cumbersome and time intensive. With gaps upward of a year between initial suspicion and diagnosis, valuable time where treatments and behavioral interventions could be applied is lost as these disorders remain undetected. Methods to quickly and accurately assess risk for these, and other, developmental disorders are necessary to streamline the process of diagnosis and provide families access to much-needed therapies sooner. Using forward feature selection, as well as undersampling and 10-fold cross-validation, we trained and tested six machine learning models on complete 65-item Social Responsiveness Scale score sheets from 2925 individuals with either ASD (n=2775) or ADHD (n=150). We found that five of the 65 behaviors measured by this screening tool were sufficient to distinguish ASD from ADHD with high accuracy (area under the curve=0.965). These results support the hypotheses that (1) machine learning can be used to discern between autism and ADHD with high accuracy and (2) this distinction can be made using a small number of commonly measured behaviors. Our findings show promise for use as an electronically administered, caregiver-directed resource for preliminary risk evaluation and/or pre-clinical screening and triage that could help to speed the diagnosis of these disorders.
View details for DOI 10.1038/tp.2015.221
View details for PubMedID 26859815
View details for PubMedCentralID PMC4872425
- Propagation of singularities around a Lagrangian submanifold of radial points Bulletin de la SMF 2015
PROPAGATION OF SINGULARITIES AROUND A LAGRANGIAN SUBMANIFOLD OF RADIAL POINTS
BULLETIN DE LA SOCIETE MATHEMATIQUE DE FRANCE
2015; 143 (4): 679-726
View details for Web of Science ID 000371649000003
- A Normal Form Around a Lagrangian Submanifold of Radial Points INTERNATIONAL MATHEMATICS RESEARCH NOTICES 2014: 4804-4821
- The Feynman propagator on perturbations of minkowski space. arXiv.org 2014
- Microlocal analysis of Lagrangian submanifolds of radial points Stanford University Thesis 2013
- Color-Permuting Automorphisms of Cayley Graphs Congressus Numerantium 2008; 190: 161-177