Zoom Link: https://github.com/giorgianb/spdhackspring2021/blob/main/bit.ly/spdhack2021

ΣΦΔ Hack Spring 2021 is ΣΦΔ's first annual machine learning hackathon. ΣΦΔ Hack Spring 2021 aims to introduce Stony Brook students to the rich and challenging field of machine learning, and develop the skills necessary to build sophisticated machine learning models on their own.
 
More info here: https://github.com/giorgianb/spdhackspring2021/blob/main/README.md
Abstract: Visual generation is a fundamental problem in computer vision and graphics, with applications ranging from 3D capture to content creation and image/video synthesis. Despite rapid progress in neural rendering and generative models, efficiency remains a key obstacle in practice: high-quality 3D reconstruction often depends on dense multi-view supervision; scalable 3D synthesis faces heavy optimization, training, and rendering costs; and modern image/video generators incur substantial computation as token grids grow with spatial resolution and temporal length.
This thesis targets efficient visual world modeling by improving sample efficiency in 3D reconstruction, representation efficiency in 3D generation, and computational efficiency in image/video synthesis. First, we improve sample efficiency for neural implicit surface reconstruction under sparse views by integrating multi-view stereo probability volumes as a geometric regularizer, enabling high-quality reconstruction from as few as three input images. Next, we introduce an explicit 3D representation for 3D generation, built from multi-view depth and RGB predictions with 3D Gaussian features, which enables the use of 2D generative priors while enforcing multi-view consistency via epipolar attention. We then address the computational bottleneck of image and video synthesis with importance-based token merging, using importance signals available during generation to preserve critical information while merging redundant tokens. Finally, we propose efficient mixed-resolution diffusion transformers via cross-resolution phase-aligned attention, aiming to improve attention stability under mixed token grids and support high-fidelity mixed-resolution generation.

Speaker: Haoyu Wu

Location: NCS120
Abstract: Many scientific and engineering challenges, such as the design of materials or molecules or the control of experimental systems, rely on the existence of fast predictive models that can evaluate potential designs or control policies. Traditionally this has been accomplished through numerical simulation; more recently data-driven machine learning methods have been applied. However, both approaches leave gaps: physical modeling can be accurate and extrapolates well to previously-unstudied conditions, but it is often computationally expensive and relies on physics approximations that may not be valid. Machine learning can generalize from massive amounts of real-world or simulation data, but suffers from physical grounding and extrapolation into new regimes, as well as in settings where large data sets do not exist.
In this talk I explore an intermediate regime, which is hybrid reduced order models: fast simplified physics approximations where some of the unknown or approximated equations are replaced with data-driven machine learning components. Examples include coarse-grained models where the full macroscopic equations cannot be derived from first-principles microscopic equations, multiscale models with unknown closure terms or sub-grid parameterization schemes, and low-order or latent dynamical systems that learn governing equations on a low-dimensional reduced state space. I discuss how such reduced systems can be identified from very limited data, much less than is often needed in traditional machine learning but at much lower time-to-solution than traditional numerical modeling. This facilitates not only system design and control but also uncertainty quantification approaches that search the space of possible equations for predictive models that can explain the data. I will focus on an example from materials science concerning the design of self-assembling block copolymer nanomaterials.

Speaker: Dr. Nathan Urban, Applied Mathematics Department, Brookhaven National Laboratory

Location: Laufer 101

Zoom: https://stonybrook.zoom.us/j/96090260834?pwd=mw8QTHbMOw9oeU9hazZeoq8bN4VIfH.1
Meeting ID: 960 9026 0834 Passcode: 374969
Title: Class visual similarity based noisy sample removal in generative Few Shot Learning
Time: Thursday, Feb 4, 11:30am - 1:00pm
Zoom:
https://stonybrook.zoom.us/j/8563646526?pwd=anJna1gzUStXNlNVSUIzdDRUSC9CUT09

Meeting ID: 856 364 6526
Passcode: 203791



Abstract:  

Over the past decade, larger datasets, hardware accelerations, and network architecture improvements have contributed to phenomenal achievements in many tasks of computer

vision. However, in the absence of large datasets, computer vision models struggle to learn

general representations which results in poor performance. Few-shot learning tries to address 

this problem by proposing models which learn from a few examples.


I first give an overall review of few-shot learning methods. I particularly focus on generative Few Shot Learning(FSL) methods, which augment the scarce categories in a dataset by generating samples for those rare categories. As the actual class distribution can be complex and lie very close to each other, the sample generated for one class can be noisy or lie close to another class.  However, none of the current FS generative methods perform any form of quality control of the generated samples.


In this work, I propose to identify and remove the generated samples that are less likely to be in the distribution of the few-shot class. Here I particularly deal with few-shot scenarios where the

prior information of the relationship between the classes based on visual  similarity is available. The main idea is to exploit these priors to better identify the unreliable generated samples.


Particularly, I have proposed two methods based on class relationship to detect noisy generated samples. In the first method, we assume that the embedding space of each class follows a Gaussian distribution.  From this assumption, I propose Gaussian Neighborhood (GN), a method to estimate how likely a generated sample is drawn from the estimated distribution of a few-shot class.  We evaluate this method on the Hematopoiesis dataset.  By simply eliminating samples based on thresholding our proposed GN scores, the few-shot  classification  performance  is  improved by 5% and 2% in five shot and one shot respectively, compared to the model trained on all generated images. 


The GN scores represent the similarity distances from the generated samples to their classes, based on the assumption that each class is a Gaussian distribution.  However, this assumption might be strict in many scenarios since the real distributions of data can be arbitrarily complex.  Thus in my second proposed method, I aim to learn such similarity distances directly from data via metric learning. I propose to train a deep-network to regress the similarity distance between a pair of samples.  This network is trained using both the class-level  visual  similarity  information  and  the  class  labels.   This method improves the 1-shot and 5-shot classification performances by 0.5% and 1% respectively, compared to GN.
























A new virtual seminar series on Games, Decisions, and Networks will start this Friday. The series aims at bringing together researchers working on foundations and applications of games theory, decision theory, and networks from computer science, control, economics and operation research. 





The advisory board for the series comprises Asu Ozdaglar (MIT), Christos Papadimitriou (Columbia), Drew Fudenberg (MIT), Eva Tardos (Cornell), Matthew O. Jackson (Stanford), Ramesh Johari (Stanford), and Tamer Başar (UIUC). 
The first talk will be given by Costantinos Daskalakis (MIT) on January 22nd at noon ET, titled Equilibrium Computation and the Foundations of Deep Learning. Upcoming speakers include


- Rakesh Vohra (Upenn)
- Sanjeev Goyal (Cambridge)
- Aaron Roth (Upenn)
- Aislinn Bohren (Upenn)
- Jason Marden (UCSB)

and more to be added!
Abstract: Large Language Models (LLMs) have transitioned from standalone prediction interfaces into integrated systems that incorporate content protection, external knowledge retrieval, and multi-step reasoning. While these functional layers expand model capabilities, they also introduce complex, inter-component dependencies that create novel and systemic security risks. This research provides a systematic deconstruction of the structural vulnerabilities emerging across these functional layers.

In this proposal, we evaluate the security boundaries of LLM systems through three pivotal dimensions:
The Content Layer: We present Watermark under Fire, revealing the inherent fragility of content-based tracing mechanisms under adaptive perturbations and highlighting the limitations of surface-level safety measures.
The Retrieval Layer: We introduce GraphRAG under Fire to examine the security of topology-aware knowledge integration. We reveal how graph-based indexing can be exploited as a structural lever for high-success poisoning attacks.
The Reasoning Layer: We detail AutoRAN, the first framework demonstrating the hijacking of internal safety reasoning in Large Reasoning Models (LRMs). This work proves that the transparency of the reasoning process itself creates a critical and exploitable attack surface.

Collectively, these studies demonstrate a systemic failure of add-on safety mechanisms in securing the broader LLM ecosystem. By identifying recurring patterns of exploitation across different system layers, this research provides the necessary foundation for transitioning from reactive patching to a more unified and architecturally-grounded approach to AI trustworthiness.

Speaker: Jiacheng Liang

Zoom: https://stonybrook.zoom.us/j/6669990420?pwd=dkY0eEw5YXpPSWo3RUE4OE1oVW90UT09&omn=97367037382
Meeting ID: 666 999 0420
Passcode: 075299

We invite faculty to deliver a 10-minute presentation during our afternoon session at the CELT Symposium on April 11, 2025. Showcase how you use emerging technology (i.e. AI, VR, etc.) to support diverse student populations and enhance learning experiences. Share your innovative strategies and inspire others!

CELT Symposium Theme: A New Era of Inclusivity and Innovation in Higher Education

https://t.e2ma.net/click/5w0gph/5wwlu4oe/9v63j6
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.