Serge Assaad

I am a scientist passionate about machine learning (ML) and its applications to impactful problems.

Through industry experience at Waymo and Amazon, I have designed scalable, fast, accurate, and robust ML models for autonomous driving, trajectory prediction, and biometric recognition. Through my Ph.D. research, I have experience in ML for causal inference and ML for digital pathology.

I'm a Ph.D. student at Duke University, working with Dr. Lawrence Carin.

I'm currently working on a few things:

  • Rotation-equivariant models for classification and trajectory prediction

  • Treatment effect estimation using deep learning

  • Computer Vision/Multiple-Instance Learning and applications to thyroid & prostate pathology


Before my Ph.D. , I completed a B.S. in Biomedical Engineering and Electrical Engineering (Mathematics Minor) at Duke.

Email: serge [dot] assaad [at] duke [dot] edu

[Linkedin] [Github] [Google Scholar] [CV]


Experience

Research Intern, Waymo (September - December 2021)

Applied Scientist Intern, Amazon (June - September 2021)

Applied Scientist Intern, Amazon (May - August 2020)

Papers





S. Assaad, C. Downey, R. Al-Rfou, N. Nayakanti, B. Sapp, "VN-Transformer: Rotation-Equivariant Attention for Vector Neurons" (preprint) [arXiv]



S. Assaad, S. Zeng, H. Pfister, F. Li, L. Carin, "Hölder Bounds for Sensitivity Analysis in Causal Reasoning" , Workshop on the Neglected Assumptions in Causal Inference, International Conference on Machine Learning (ICML), 2021 [paper][poster]


D. Dov, S. Assaad, A. Syedibrahim, J. Bell, J. Huang, J. Madden, R. Bentley, S. McCall, R. Henao, L. Carin, W. Foo, "A hybrid human-machine learning approach to screening prostate biopsies can improve clinical efficiency without compromising diagnostic accuracy", Archives of Pathology & Laboratory Medicine, 2021 [paper]



D. Dov, S.Z. Kovalsky, Q. Feng, S. Assaad, J. Cohen, J. Bell, R. Henao, L. Carin, D.E. Range, ``Use of Machine-Learning-Based Software for the Screening of Thyroid Cytopathology Whole Slide Images'', Archives of Pathology & Laboratory Medicine, 2021 [paper]


S. Assaad, S. Zeng, C. Tao, S. Datta, N. Mehta, R. Henao, F. Li, L. Carin, "Counterfactual Representation Learning with Balancing Weights" , International Conference on Artificial Intelligence and Statistics (AISTATS), 2021 [paper][poster][talk]


P. Chapfuwa, S. Assaad, S. Zeng, M. Pencina, L. Carin. R. Henao, "Enabling Counterfactual Survival Analysis with Balanced Representations", ACM Conference on Health, Inference, and Learning (ACM-CHIL), 2021 [paper][talk]

D. Dov, S. Assaad, S. Si, R. Wang, H. Xu, S. Kovalsky, J. Bell, D. Range, J. Cohen, R. Henao, L. Carin, "Affinitention Nets: Kernel Perspective on Attention Architectures for Set Classification with Applications to Medical Text and Images", ACM Conference on Health, Inference, and Learning (ACM-CHIL), 2021 [paper][talk]


D. Dov, S.Z. Kovalsky, S. Assaad, J. Cohen, D.E. Range, A.A. Pendse, R. Henao, L. Carin, "Weakly Supervised Instance Learning for Thyroid Malignancy Prediction from Whole Slide Cytopathology Images", Medical Image Analysis, 2020 [paper]


S. Zeng, S. Assaad, C. Tao, S. Datta, L. Carin, F. Li, "Double Robust Representation Learning for Counterfactual Prediction" (preprint) [arXiv]


Professional Services

Reviewer for NeurIPS (2022), ICML (2022), AISTATS (2021), Neural Networks (2021)

Talks/Posters

Research talks

  • Rotation-equivariant Transformers for Behavior Prediction, Waymo Research (2021)

  • Hölder Bounds for Sensitivity Analysis in Causal Reasoning, ICML Workshop on Neglected Assumptions in Causal Inference (2021) [poster]

  • Quantization and Distillation for Biometric Recognition, Amazon Research (2020)

  • Counterfactual Representation Learning with Balancing Weights, AISTATS (2021) [poster]

  • Machine Learning of Continuous Wave Form Doppler Signals, Society for Clinical Vascular Surgery (2018) [link]


Tutorials

  • Introduction to PyTorch for machine learning (350+ attendees), Duke Winter Breakaway (2021) [link]

  • Introduction to PyTorch for deep learning, Duke +Data Science (2020) [link]

  • Variational inference, Duke Datathon (2019) [code] [link]

  • Crash course in do-calculus for causal inference (2019) [slides]

  • Image classification with Tensorflow, Duke Datathon (2018)

  • Keynote speech, Duke International Orientation (2018)


Reading group

  • Zheng et al., "Copula-based Sensitivity Analysis for Multi-Treatment Causal Inference with Unobserved Confounding" (2021) [slides]

  • Johansson et al., "Learning Weighted Representations for Generalization Across Designs" (2018) [slides]

  • Saito & Yasui, "Counterfactual Cross-Validation: Stable Model Selection Procedure for Causal Inference Models" (2020) [slides]

  • A. D'Amour, "On Multi-Cause Causal Inference with Unobserved Confounding: Counterexamples, Impossibility, and Alternatives" (2019) [slides]

  • Louizos et al., "Causal Effect Inference with Deep Latent-Variable Models" (2017) [slides]

  • Johansson et al., "Learning Representations for Counterfactual Inference" (2016) [slides]

  • Shalit et al., "Estimating Individual Treatment Effect: generalization bounds and algorithms" (2017) [slides]

  • Kiros et al., "Skip-thought vectors" (2015) [slides]

  • Cer et al., "Universal Sentence Encoder" (2018) [slides]

Teaching

Graduate TA Performance & Technology (2021), Vector Space Methods (2020), Data Science Seminar (2019)

Undergraduate TA Calculus (I, II, III) (2015), Linear Algebra (2015), Probability (2015), Data & Decision Sciences (2018), Introduction to Circuits (2017)

Tutor Boeing Grand Challenges Fellow (2018), NCAA Student-Athlete Tutor (2016)

Misc. Interests

Salsa dancing (member of Duke Sabrosura), Coffee, Tennis, Lebanese cuisine