Abstract: This dissertation addresses the methodological disconnect between Natural Language Processing (NLP) and human-centric analysis by shifting the unit of analysis from document to human behavior in two broad respects: (i) time-ordering: modeling documents as sequential person-indexed behavioral observations, and (ii) person-level semantics: evaluation and explainability of models by their latent structure of psychological constructs rather than just its predictive accuracy against narrow proxy measures. First, we consider the most basic implication of language as a person's behaviors when measuring their psychological constructs: relationship between language sample size and model's predictive performance. We empirically show that the state-of-the-art transformers are often over-parameterized for typical NLP dataset sizes and can be reduced in dimensionality without performance loss. Establishing the author as the unit of analysis naturally allows us to treat their behavior as a time-ordered sequence. Second, we introduce a longitudinal evaluation framework that establishes ecologically valid evaluation settings, namely, cross-sectional and prospective generalization, and separates error measurement of the model into within-person dynamics and between-person differences. We demonstrate that traditional NLP evaluations based on random document splits can yield reversed conclusions under ecologically valid generalization settings. To address this, we develop models that capture the trajectory of mental states (e.g., mood shifts) rather than static traits. Third, moving into person-level semantics, we evaluate the latent structure of large language models using a novel machine behavior analytic framework. We find that while GPT-4 achieves high predictive correlation with self-reports, its latent symptoms structure diverges from clinical understanding. Finally, we propose a method for modeling multidimensional behaviors, embedding concurrent behavioral signals alongside language to predict future states. Taken together, this work suggests that operationalizing language as behavior advances NLP methods into a rigorous instrument for valid psychological inquiry.

Speaker: Adithya Ganesan

Location: Join Zoom Meeting (ID: 99021939129, Passcode: 569493)

Subject: RADIOLOGY GRAND ROUNDS CT Colonography: An Effective Test for Colorectal Cancer Screening- Judy Yee, M.D.
When: Wednesday, May 12, 2021 12:00 PM-1:00 PM (UTC-05:00) Eastern Time (US & Canada).
Where: JOIN ZOOM MEETING

 

Judy Yee, MD

Chair, Department of Radiology

Professor, Department of Radiology

Abdominal Imaging

 

Join Zoom Meeting

https://einsteinmed.zoom.us/j/97782190723?pwd=clMzMys2SlZjZzJId1hUNzMyVUQ2UT09

 

Meeting ID: 977 8219 0723

Passcode: 101083

The Office for Research and Innovation at Stony Brook University invites you to attend the inaugural Wolf Den, an evening designed to bring together members of the regional innovation and entrepreneurial ecosystem.

Meet investors, researchers, startup founders, and business leaders to exchange ideas, foster collaboration, and strengthen connections that drive technology development and economic growth across Long Island.

Agenda

4:30 - 5:00 PM | Grab some cheer & mingle
5:00 - 5:40 PM | Welcome remarks and AI Panel
5:40 - 6:00PM | Featured lightning pitches
6:00 - 7:00 PM | Food, drinks and great conversations!

Attendees will have the opportunity to learn more about Stony Brook's entrepreneurship ecosystem, hear company pitches from emerging startups, and engage in meaningful networking with innovators, investors and community partners.

Refreshments will be served. Registration is required.

In partnership with Accelerate Long Island.

https://www.stonybrook.edu/commcms/innovation/_events/wolfden.php

Abstract: The remarkable success of large foundational models, such as LLMs and diffusion models, is built on their learning over vast amounts of static data from the Internet. However, human learning and problem-solving are fundamentally interactive processes--humans learn by engaging with their environment, tools, search engine, and feedback loops, iteratively refining their understanding and decisions. This gap between the interactivity of human learning and the static nature of model training raises a critical question: how can we imbue foundational models with the capacity for meaningful interaction?

In this talk, I will explore methods to enhance foundational models by incorporating interaction with the external environment. I will discuss strategies such as leveraging external tools, compilers, function calls to provide dynamic feedback to enhance foundation models. By drawing inspiration from human's interactive learning processes, I demonstrate how interaction-driven learning can lead to models that are not only more accurate but also more adaptable to real-world applications.

This work bridges the gap between static training paradigms and the dynamic, iterative nature of human intelligence, paving the way for a new generation of interactive AI systems.

Bio: Wenhu Chen has been an assistant professor at the Computer Science Department in University of Waterloo and Vector Institute since 2022. He obtained the Canada CIFAR AI Chair Award in 2022 and CIFAR Catalyst Award in 2024. He has worked for Google Deepmind as a part-time research scientist since 2021. Before that, he obtained his PhD from the University of California, Santa Barbara under the supervision of William Wang and Xifeng Yan. His research interest lies in natural language processing, deep learning and multimodal learning. He aims to design models to handle complex reasoning scenarios like math problem-solving, structure knowledge grounding, etc. He is also interested in building more powerful multimodal models to bridge different modalities. He received the Area Chair Award in AACL 2023, the Best Paper Honorable Mention in WACV 2021, the Best Paper Finalist in CVPR 2024, and the UCSB CS Outstanding Dissertation Award in 2021.
Abstract:
People shift their visual attention to gather and prioritize information from their surroundings, helping them navigate complex environments. Understanding these attentional shifts involves decoding the features that guide where attention is directed (spatial areas of focus) and when attention shifts (timing). Decoding these processes can aid applications from interface design to medical diagnosis. However, prior models have not fully explored the underlying factors addressing these aspects. In this dissertation, we study the factors that guide visual attention across diverse image types, spanning natural images, graphic design documents, and whole slide images (WSIs) of cancer tissues, while also predicting visual attention based on these factors.
First, we propose a method to quantify object recognition uncertainty as a factor influencing spatio-temporal attention (where and when) in natural images. We found that it plays a larger role than bottom-up saliency in guiding visual attention. Second, we analyze graphic design documents such as webpages, comics, posters, mobile UIs, etc., which differ from natural images in that they are designed to convey specific messages or elicit desired viewer response. We propose a unified and interpretable deep learning model that predicts both static and dynamic visual attention behavior (addressing where and when) by integrating document layout and content saliency as factors, enhancing attention prediction performance. Finally, in the domain of digital pathology, we investigate pathologists' attention during their examination of giga-pixel WSIs of prostate cancer with an objective to aid in the development of computer-assisted pathology training and clinical decision support systems. Using a digital microscope interface, we collected the largest known dataset of pathologist attention, which allows us to study the factors that guide their spatial and temporal attention patterns (where and when) and develop predictive models. Our study explores key factors guiding their attention, including magnification, slide staining, the nature of the diagnostic task, and their expertise. Motivated by this analysis, we propose deep learning models to solve two tasks: 1) predicting pathologist attention via spatial (heatmaps) and spatio-temporal (scanpaths) models, and 2) inferring pathologist expertise level, both essential technical components towards developing an AI-assisted pathology training pipeline.

Speaker:
Souradeep Chakraborty

Location: New Computer Science Bldg., Room 220

Zoom Link: https://stonybrook.zoom.us/j/9755288447?pwd=TW95T2xqOUZjRnlqcnVFcUQvN0JMdz09
Meeting ID: 975 528 8447
Passcode: 338037


The Program in Writing and Rhetoric
Invites you to
A Rhetorical/Deliberative Framework for AI Language Model Alignment
featuring
Prof Zoltan Majdik Professor
North Dakota State University
In this talk, Prof. Majdik proposes a framework for aligning LLMs with values grounded in the norms of rhetorical culture and deliberative democracy. Alongside long-standing AI alignment value targets like safety and transparency, this AI alignment framework assesses to what extent a language model exhibits human and humane values that foster communicative engagement, and it codifies approaches to tuning existing models to better align with such values.

Location: Humanities 1008
The Empirical Methods in Natural Language Processing (EMNLP) conference is a premier international academic conference in the field of artificial intelligence and natural language processing (NLP). Organized annually by the Association for Computational Linguistics (ACL) special interest group on linguistic data (SIGDAT), it focuses on research that uses empirical methods to solve language processing problems.

For more information, and registration, visit the official website.
Defending Software Systems from Cyber Attack Campaigns Presented by R. Sekar The DNC hack of 2016, the Equifax breach of 2017, and the spate of ransomware campaigns in 2019 demonstrate the formidable challenges we face in securing our network and software systems against highly stealthy and sophisticated adversaries. In this talk, I will describe two avenues of research we have been pursuing to help tilt the table against such powerful adversaries. The first is software hardening techniques that make software vulnerabilities harder to exploit. To maximize their applicability and ease of use, our techniques are implemented into compilers, or they directly transform binary code. I will outline some of the exciting new developments we have had in this area over the years, including randomization, memory safety, information-flow tracking, control-flow integrity, and code-pointer integrity. We complement this first line of defense with techniques for analyzing and understanding attack campaigns that manage to slip past all deployed defenses. Our techniques can sift through logs consisting of hundreds of millions of events to zoom in on attack activity that may span just a few hundred events. I will describe our experience in mapping out several DARPA-sponsored red team attack campaigns.