Abstract: Pre-trained diffusion and flow matching models have made visual generation remarkably powerful, enabling high-fidelity synthesis of images and videos from natural language prompts. However, their behavior is still largely dictated by the pre-training data distribution and likelihood objective, which do not directly encode downstream desiderata such as fine-grained semantic alignment, controllability, or realism. This gap motivates post-training: starting from a base generator and further optimizing it with additional supervision signals derived from human or reward model preferences.This work presents post-training for visual generative models through two complementary case studies. First, Hummingbird addresses the problem of fine-grained contextual alignment in image-text-to-image generation. We introduce a multimodal context evaluator that scores the consistency between rich contextual descriptions and generated images, capturing fine-grained alignment beyond global CLIP similarity. By directly backpropagating these differentiable rewards through the diffusion sampler, Hummingbird substantially improves semantic faithfulness while preserving high visual quality.
Second, PISCES tackles post-training for text-to-video generation, where alignment is inherently semantic-spatio-temporal. We show that naive VLM-based rewards suffer from distributional mismatch and token-level misalignment, leading to reward hacking and suboptimal optimization. PISCES introduces a bi-objective, Optimal Transport (OT)-aligned reward module: distributional OT using Neural Optimal Transport to align text and video embedding distributions, and discrete, partial OT over a spatio-temporal cost matrix to capture semantic alignment at the token level. These rewards are integrated into both direct backpropagation and GRPO-style optimization to post-train state-of-the-art text-to-video generators. Together, Hummingbird and PISCES provide a unified view of how carefully designed visual reward models, coupled with OT-based representation alignment, can reliably improve the downstream behavior of pre-trained image and video generators.

Speaker: Minh Quan Le

Location: NCS 220

Zoom: https://stonybrook.zoom.us/j/94798224254?pwd=CFraer25qnpORbJ14aAVHRwaSJOjJM.1
Abstract: Anxiety disorders are characterized by persistent and excessive form of fear and worry that interferes with daily functioning, distinguishing it from the adaptive anxiety that helps individuals respond to challenges. Despite affecting millions worldwide and costing a significant public health burden, anxiety disorders still remain underdiagnosed than actual prevalence due to lack of understanding and stigmatization. Leveraging machine learning (ML) and natural language processing (NLP) approaches can help bridge this gap by enabling scalable and accessible mental health assessments, offering a data-driven understanding of anxiety from individual and societal perspectives, and shedding light on societal stigmas toward mental health conditions. At the same time, advancing ML and NLP techniques for anxiety research presents unique technical challenges, such as effectively modeling linguistic markers of anxiety and ensuring interpretability in mental health predictions.

This dissertation investigates anxiety from both individual and societal perspectives using artificial intelligence. First, we explore individual manifestations of anxiety through three methodological advancements: (1) integrating contextual and discourse-level embeddings to improve language-based anxiety prediction using Facebook posts and selfreported surveys; (2) enhancing cognitive dissonance detection in Twitter dataset with transfer learning and active learning; and (3) developing longitudinal representation learning approaches that achieve both predictive utility and interpretability of adolescent psychopathology. Finally, we extended our analysis to societal dimension of anxiety by identifying and categorizing social norms expressed in Reddit and Twitter posts and examining their associations with anxiety. By combining data-driven methods with psychological insights, this work studies anxiety from various angles - capturing both individual experiences and societal influences - offering a step toward a more comprehensive understanding of its causes and manifestations.

Speaker: Swanie Juhng

https://stonybrook.zoom.us/j/98905245099?pwd=M7rI7aNfNio281qyebEUdNPBcSiK7Y.1
Abstract: Neurosymbolic AI, as the combination of deep learning and knowledge representation/reasoning methods, is currently receiving significant attention due to its promise to both overcome apparent limitations of pure deep learning systems, and to address the knowledge acquisition bottleneck. This presentation will focus on some recent reseach advances made at Kansas State University related to Semantic Web / knowledge graphs and neurosymbolic AI, with particular emphasis on the use of symbolic AI methods for understanding hidden neuron activations, and on using large language models for knowledge graph and ontology engineering.

Short bio: Pascal Hitzler is University Distinguished Professor and endowed Lloyd T. Smith Creativity in Engineering Chair at the Department of Computer Science at Kansas State University, one of the Directors of the Institute for Digital Agriculture and Advanced Analytics (ID3A), and Director of the Center for Artificial Intelligence and Data Science (CAIDS). His research record lists over 400 publications in such diverse areas as neurosymbolic artificial intelligence, semantic web, knowledge graphs, knowledge representation and reasoning, denotational semantics, and set-theoretic topology. He was founding Editor-in-chief of the Semantic Web journal, the leading journal in the field, and is founding Editor-in-chief of the new Neurosymbolic Artificial Intelligence journal. He is co-author of the W3C Recommendation OWL 2 Primer, and of the book Foundations of Semantic Web Technologies by CRC Press, 2010, which was named as one out of seven Outstanding Academic Titles 2010 in Information and Computer Science by the American Library Association's Choice Magazine, and has translations into German and Chinese. He is a founding steering committee member of the Neural-Symbolic Learning and Reasoning Association that runs the annual International Conference on Neurosymbolic Learning and Reasoning (NeSy). For more information about him, see http://www.pascal-hitzler.de.

Location: NCS 120

Zoom: https://stonybrook.zoom.us/j/94657223635?pwd=FHtIvIupDAbQTCKzpDF3kTg6xRZOtj.1&jst=2

Abstract: Current Chain-of-Thought (CoT) verification methods predict reasoning correctness based on outputs (black-box) or activations (gray-box), but offer limited insight into \textit{why} a computation fails. We introduce a white-box method: \textbf{Circuit-based Reasoning Verification (CRV)}. We hypothesize that attribution graphs of correct CoT steps, viewed as \textit{execution traces} of the model's latent reasoning circuits, possess distinct structural fingerprints from those of incorrect steps. By training a classifier on structural features of these graphs, we show that these traces contain a powerful signal of reasoning errors. Our white-box approach yields novel scientific insights unattainable by other methods. (1) We demonstrate that structural signatures of error are highly predictive, establishing the viability of verifying reasoning directly via its computational graph. (2) We find these signatures to be highly domain-specific, revealing that failures in different reasoning tasks manifest as distinct computational patterns. (3) We provide evidence that these signatures are not merely correlational; by using our analysis to guide targeted interventions on individual transcoder features, we successfully correct the model's faulty reasoning. Our work shows that, by scrutinizing a model's computational process, we can move from simple error detection to a deeper, causal understanding of LLM reasoning.

Speaker: Xianjun Yang

Location: Old Computer Science Building - CS2311
The INS (International Neuroethics Society) AI and Consciousness Affinity Group is hosting a talk titled Bringing Trustworthiness in Generative AI and Agentic AI Using Thought Knowledge Graphs featuring speaker Manas Gaur, a computer scientist at UMBC.
The talk will examine the interplay between Thought Knowledge Graphs (TKGs) and how they can form more trustworthy and reasoning-based responses in AI. They will also discuss introducing novel methods on implementing TKGs and their overall impact on creating more trustworthy AI systems.
The talk will be held online via Zoom on Monday, December 2 at 1:00pm (EST).
Register to attend.
The Institute for AI-Driven Discovery and Innovation hosts Dr. Mary
Simoni for a talk on her music and its intersection with AI, as part
of the Music and AI Seminars series.

The event will be held on Thursday, December 10, 2020, at 3:00 PM.

Abstract: Mary Simoni, Dean of Humanities, Arts & Social Sciences at
Rensselaer Polytechnic Institute will discuss her research in the use
of computer algorithms and technology in the composition and
performance of music. The talk will feature compositions inspired by
Augmented Transition Networks (ATNs), employ motion tracking to
control synthesis parameters, and a work in progress that employs
machine learning using training data that juxtaposes classical music
with COVID-19. During this talk, participants will be introduced to
several technologies that support music information retrieval, machine
learning, and algorithmic composition such as jSymbolic, Weka, and
Common Music.

Zoom details below:
https://stonybrook.zoom.us/j/98236706900?pwd=bDFEZFZtaHBWU0cyL0wxK3UrdUpIdz09
Meeting ID: 982 3670 6900
Passcode: 133945  

This course is designed to help you approach AI with clarity and intention. Explore how to ask better questions, apply critical judgment to AI-generated outputs, and use these tools to support--not replace--human decision-making. With the right prompts, a thoughtful lens, and a focus on impact, AI can help reduce friction and free you to focus on what matters most: people, purpose, and leadership.

Audience: Any current or aspiring leader who currently uses AI and is interested in bringing their skills to the next level.

Register here.


Are you concerned about AI issues with your asynchronous online courses? Is your fully online course vulnerable to AI plagiarism? Do you want to engage your online students using AI? Discover the future of education with our AI-powered solutions designed specifically for online asynchronous courses. This innovative approach uses artificial intelligence to transform the way courses are delivered, making learning more personalized, engaging, and effective.

Register here.