paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

arXiv

Why does AI unlock new possibilities in STEM education? A Bibliometric Analysis of Trends and Future Agenda

Jesse Yusuf Chan, Mengyao Chen, Yang Hong, Ziyun Song, Haoming Wang, Xianlong Xu

STEM education faces challenges in personalization and interdisciplinary integration. AI technology has brought new possibilities, but the mechanisms by which AI reshapes the STEM education ecosystem require systematic investigation. This study employs bibliometric methods to analyze 242 publications from 2015-2025, constructing knowledge maps to reveal the evolutionary trajectory. The findings show that the field has transformed from intelligent tutoring systems to inquiry-based learning and computational thinking cultivation driven by LLMs. AI's key contribution lies in providing intelligent scaffolding that lowers the threshold for understanding knowledge. In this sense, AI is a core driving force promoting its shift from knowledge transmission to capability development.

arXiv

Matching Matters: A Fair Quality-Efficiency Benchmark for Command-Line Agents

Han Chi, Jiaxin Qi, Yan Cui, Baisheng Lai, Jianqiang Huang

Rapid advances in large language models have improved the task-solving capabilities of command-line-interface (CLI)-based agents, whose CLIs determine how models invoke tools, maintain interaction history, and recover from failures. Consequently, effective matching between CLIs and LLMs has become essential. However, existing agent benchmarks largely emphasize success rate while overlooking practical objectives such as cost and efficiency, as well as the selection of LM-CLI combinations, all of which are critical in real-world deployment. We therefore introduce AgentMeter, a quality-efficiency benchmark with a new metric, the AgentMeter Score (AMS), that jointly characterizes task quality, budget sensitivity, and resource-intensive zero-reward execution, enabling a more complete assessment of deployed LM-CLI pairs. Furthermore, collected task descriptions may inadvertently favor LM-CLI pairs that are particularly compatible with their wording and structure, causing evaluation results to reflect description-specific advantages rather than general task-solving capability. We therefore propose AgentMeter-Opt, a trajectory-grounded optimization framework that constructs pair-adapted, task-preserving description variants to build a fairer evaluation set across LM-CLI pairs. Extensive experiments show that no CLI is universally optimal across language models and that task success, execution cost, and AMS identify different competitive configurations. Results on AgentMeter-Opt further reveal that task-preserving description changes affect LM-CLI pairs unevenly and can alter their relative ordering across valid description conditions. Together, AgentMeter and AgentMeter-Opt provide a practical foundation for fair and deployment-relevant evaluation of command-line agents.

arXiv

CASA: Classification Augmented with Safety Attention for Robust Multimodal Alignment

Anurag Kumar, Raghuveer Peri, Jon Burnsky, Alexandru Nelus, Rohit Paturi, Srikanth Vishnubhotla, Yanjun Qi

Multimodal large-language models (MLLMs) often experience degraded safety alignment when harmful queries exploit cross-modal interactions. Models aligned on text alone show a higher rate of successful attacks when extended to two or more modalities. We propose a simple conditional decoding strategy, CASA (Classification Augmented with Safety Attention) that uses internal representations of MLLMs to predict a binary safety token before response generation. The classification is guided by a safety attention mechanism that computes attention between a dedicated query prompt and the input to scale the classification logits, sharpening the model's ability to detect malicious queries. Our design ensures robust safety alignment without relying on any external classifier or auxiliary head, and without modality-specific safety fine-tuning. Unlike prior safety-decoding methods that attach an external classifier or moderation head, or rescale every generated token with a reward model, CASA makes a single, internal, attention-gated safety decision over the shared multimodal representation, which is precisely what lets safety learned from text-only training transfer to image and audio inputs. Compared against 4 baseline models across 7 attack types spanning text, image, and audio modalities, evaluated on benchmarks such as MM-SafetyBench, JailbreakV-28k, and adversarial audio tests, CASA lowers the average attack success rate by >90% relative when paired with a standard safety prompt (applied to all baselines for fair comparison), and by 54--61% using its internal safety mechanism alone. Our empirical evaluations also show that CASA~maintains strong utility on benign inputs, validated through both automated and human evaluations (via 13 trained annotators). Together, these results highlight CASA as a simple and generalizable framework to improve multimodal LLM safety.

arXiv

PrivacyPeek: Auditing What LLM-Based Agents Acquire, Not Just What They Say

Mingxuan Zhang, Jiahui Han, Dadi Guo, Songze Li, Guanchu Wang, Na Zou, Dongrui Liu, Xia Hu

LLM-based agents are rapidly advancing, autonomously invoking external tools to complete multi-step tasks for users. However, agents often acquire more sensitive information than the task requires. Existing privacy benchmarks audit what the agent's response or outgoing actions disclose, but overlook the acquisition stage where data first enters the agent's context. The over-acquired information is then one careless action or one attack away from an outright leak. To assess its prevalence, we introduce \emph{PrivacyPeek}, a benchmark for evaluating acquisition-stage privacy leakage of LLM-based agents, with $1{,}182$ cases across $7$ acquisition behaviours and $16$ application domains. Specifically, \emph{Acquisition Inspection} examines the agent's tool-call trajectory, both the tools it invokes and the data it receives, to detect when it acquires sensitive information beyond the task scope. \emph{Probe Elicitation} then issues a follow-up probe and measures how readily an attacker could elicit sensitive information the agent acquired but did not disclose. Our experiments on $10$ LLM-based agents across $4$ model families show that the unnecessary acquisition of sensitive information is widespread. In addition, we observe a correlation between the task-completion capability and acquisition-stage leakage. Prompt-level defences reduce only a small fraction of acquisition-stage leakage, leaving the majority unmitigated. These results make auditing acquisition-stage privacy both urgent and necessary. Our dataset and code are available at https://github.com/Xuan269/PrivacyPeek-Resource.

arXiv

Domain-Gated Latent Diffusion: Generative Inverse Design of HMX-Class Energetic Materials with First-Principles Validation

Yehudit Aperstein, Alexander Apartsin

Energetic materials power mining, demolition, propulsion and airbags, yet today's compounds were designed decades ago. A successor must combine high energy release, low sensitivity to accidental initiation and a practical synthesis route, found within an astronomically large molecular space. Generative models are the natural search tool, but their training data are mostly untrustworthy: of approximately 66,000 molecules with recorded properties, only approximately 3,000 were measured or computed from first principles. Models trained on all of them imitate the rough estimates and propose molecules that collapse under real physics. We introduce Domain-Gated Latent Diffusion (DGLD), a diffusion model that treats data reliability as an explicit design parameter: labels are sorted into four trust tiers, and only trustworthy ones steer generation, while the unreliable majority still teaches the model what a plausible molecule looks like. Learned controls tune performance, safety and viability independently, and every proposal passes a four-stage screen ending in a quantum-chemical DFT audit. DGLD proposes 10 molecules unknown to PubChem that survive this screen. The best, 3,4,5-trinitro-1,2-isoxazole, matches the benchmark explosives HMX and PETN in calculated detonation performance, is unlike molecules in its training set, and has a four-step synthesis route. Trust gating is chemistry-independent and can be applied wherever abundant weak data surround a reliable core.

arXiv

Behind EvoMap: Characterizing a Self-Evolving Agent-to-Agent Collaboration Network

Qiming Ye, Peixian Zhang, Yupeng He, Zifan Peng, Gareth Tyson

Agent-to-Agent (A2A) networks enable autonomous AI agents to collaborate by sharing reusable problem-solving instructions. However, how these decentralized ecosystems operate in practice remains largely unexplored. We present the first large-scale empirical study of EvoMap, a prominent A2A collaboration network. By analyzing over 1.5M assets and 128K agents, we show how design choices that prioritize scalable growth introduce trade-offs in reusability, evolution, and auditability. First, EvoMap's credit economy rewards agents for publishing valuable assets. Although this design encourages participation at scale, rewards are tied primarily to publication rather than adoption. This leads agents to mass-produce assets to accumulate credits. As a result, 98% of assets are never reused, while rewards become highly concentrated among a small fraction of agents. Second, EvoMap employs an algorithm (referred to as GDI) to score and rank the quality of these shared assets. We demonstrate that this scoring system is flawed: rather than measuring objective performance, an asset's rank is heavily dictated by unverified, self-reported metadata (e.g., claimed lines of code modified). This allows agents to trivially manipulate their asset's scores. Finally, EvoMap relies on agents to provide local execution logs as evidence that uploaded assets function correctly. Because these validations are not independently verified, over 84% of approved assets bypass quality checks using vacuous tests (e.g., console$.$log()). Our findings show that future A2A collaboration networks cannot rely on unverified self-reporting alone. Scalable collaboration requires mechanisms that balance open participation with verifiable execution and trustworthy evaluation.

arXiv

To Call or Not to Call: A Framework to Assess and Optimize LLM Tool Calling

Qinyuan Wu, Soumi Das, Mahsa Amani, Arijit Nag, Seungeon Lee, Krishna P. Gummadi, Abhilasha Ravichander, Muhammad Bilal Zafar

Agentic AI architectures augment LLMs with external tools, unlocking strong capabilities but potentially incurring substantial costs. Moreover, tool use is not always beneficial: redundant or low-utility calls can even harm task performance. Effective tool use, therefore, hinges on a core LLM decision: whether to call or not call a tool when performing a task. We introduce a principled framework inspired by decision-making theory to understand tool-use decisions along three key factors: necessity, utility, and affordability. Our analysis combines two complementary lenses: a normative perspective that infers true need and utility for optimal tool calls, and a descriptive perspective that infers the model's self-perceived need and utility from their observed behaviors. We evaluate six open models and a proprietary OpenAI model across native and customized harnesses, two tools, and six tasks. Models' perceived need and utility remain misaligned with their true values, particularly under budget constraints. This misalignment produces both costly overuse and performance-degrading calls. To improve the tool decisions, we train lightweight latent estimators of need (LNEs) from model hidden states. LNEs generally predict true need more accurately than model self-reports and improve budgeted tool allocation across model scales and tool types. Code and dataset available at https://github.com/QinyuanWu0710/ToCall_or_NotToCall.

arXiv

Text Steganography with Dynamic Codebook and Multimodal Large Language Model

Jianxin Gao, Ruohan Lei, Wanli Peng

With the popularity of the large language models (LLMs), text steganography has achieved remarkable performance. However, existing methods still have some issues: (1) For the white-box paradigm, this steganography behavior is prone to exposure due to sharing the off-the-shelf language model between Alice and Bob. (2) For the black-box paradigm, these methods lack flexibility and practicality since Alice and Bob should share the fixed codebook while sharing a specific extraction prompt for each steganographic sentence. In order to improve the security and practicality, we introduce a black-box text steganography with a dynamic codebook and multimodal large language model. Specifically, we first construct a dynamic codebook via some shared session configuration and a multimodal large language model. Then an encrypted steganographic mapping is designed to embed secret messages during the steganographic text generation. Furthermore, we introduce a feedback optimization mechanism based on reject sampling to ensure accurate extraction of secret messages. Experimental results show that the proposed method outperforms existing white-box text steganography methods in terms of embedding capacity and text quality. Meanwhile, the proposed method has achieved better practicality and flexibility than the existing black-box paradigm in some popular online social networks.

arXiv

Robust inference using density-powered Stein operators

Shinto Eguchi

We introduce a density-power weighted variant of the Stein operator, called the $γ$-Stein operator, for robust inference with unnormalized probability models. The operator is motivated by the first variation of the $γ$-divergence under infinitesimal escort transport and weights the usual Stein field by a positive power of the model density. This weighting down-weights observations in low model-density regions, providing a principled robustness mechanism while retaining the normalizing-constant-free structure of score matching. We develop the resulting $γ$-score matching estimating equations and discuss their non-integrable, generalized-method-of-moments character. We further study two extensions: a $γ$-kernelized Stein discrepancy, interpreted as a robust diagnostic or contaminated-null goodness-of-fit procedure, and $γ$-Stein variational gradient descent for robust posterior approximation. Numerical examples on directional, mixture, and quartic-potential models illustrate the robustness--efficiency trade-off: positive $γ$ can stabilize inference under targeted contamination, whereas $γ=0$ remains preferable under clean well-specified models.

arXiv

Free Denoising Diffusion Models

Swagatam Das

We develop a free-probabilistic framework for denoising diffusion, in which the data is a self-adjoint operator and its law a spectral distribution. The forward process is the free Ornstein--Uhlenbeck diffusion, whose spectral marginals solve a nonlocal Fokker--Planck equation of complex Burgers type carried by the Hilbert transform. Organising the analysis around the $1$-convexity of the free energy along free Wasserstein geodesics, we recover, in the form required by the diffusion schedule, the free logarithmic Sobolev, Talagrand and HWI inequalities of Biane, Hiai--Petz--Ueda and Ledoux--Popescu, together with a dissipation identity and the convergence of a free Jordan--Kinderlehrer--Otto scheme. Specialising Dabrowski's time reversal of free diffusions, we obtain the reverse-time equation, whose drift involves Voiculescu's conjugate variable; the latter is square-integrable at every positive time, with an explicit bound along the schedule. A free Tweedie identity yields a consistent score-matching objective and a finite-dimensional algorithm. For operator-valued volatility, we compute the spectral velocity field, match it to a Hermitian matrix model, and show that the resulting diffusion admits neither a spectral Lamperti reduction nor a Wasserstein gradient-flow structure unless the coefficient is constant. We further give a threshold for the survival of a spectral gap along the flow, in terms of the second-order Cauchy kernel of the initial law, and report numerical experiments.

arXiv

TRU: Targeted Reverse Update for Efficient Multimodal Recommendation Unlearning

Zhanting Zhou, Kahou Tam, Ziqiang Zheng, Zeyu Ma, Yang Yang

Multimodal recommendation systems (MRS) jointly model user-item interaction graphs and rich item content, but this tight coupling makes user data difficult to remove once learned. Approximate machine unlearning offers an efficient alternative to full retraining, yet current MRS unlearning applies reverse updates largely uniformly across model components. We show that this uniform treatment is misaligned with modern MRS: deleted-data influence is distributed unevenly across \textit{ranking behavior}, \textit{modality branches}, and \textit{model modules}. This non-uniformity gives rise to three bottlenecks in MRS unlearning: target-item persistence in the collaborative graph, modality imbalance across feature branches, and concentrated module-level sensitivity in the parameter space. To address this mismatch, we propose \textbf{targeted reverse update} (TRU), a plug-and-play unlearning framework for MRS. Instead of applying a uniform global reversal, TRU performs three coordinated interventions across the model hierarchy: a ranking fusion gate to suppress residual target-item influence in ranking, branch-wise modality scaling to preserve retained multimodal representations, and capacity-aware parameter-group selection to localize reverse updates to deletion-sensitive modules. Across two backbones, three datasets, and three unlearning regimes, TRU achieves a stronger retain--forget trade-off than MMRecUn in most settings. In two challenging user-level cases, TRU also attains favorable operating points among all evaluated baselines. Security audits report the lowest MIA BalAcc and a tie for the lowest ASR among approximate methods in both audited settings, while wall-clock trajectories show earlier convergence to favorable retain--forget regions.

arXiv

SimMOF: AI agent for Automated MOF Simulations

Jaewoong Lee, Taeun Bae, Jihan Kim

Metal-organic frameworks (MOFs) offer a vast design space, and as such, computational simulations play a critical role in predicting their structural and physicochemical properties. However, MOF simulations remain difficult to access because reliable analysis require expert decisions for workflow construction, parameter selection, tool interoperability, and the preparation of computational ready structures. Here, we introduce SimMOF, a large language model based multi agent framework that automates end-to-end MOF simulation workflows from natural language queries. SimMOF translates user requests into dependency aware plans, generates runnable inputs, orchestrates multiple agents to execute simulations, and summarizes results with analysis aligned to the user query. Through representative case studies, we show that SimMOF enables adaptive and cognitively autonomous workflows that reflect the iterative and decision driven behavior of human researchers and as such provides a scalable foundation for data driven MOF research.

arXiv

Using Reinforcement Learning to Optimize the Global and Local Crossing Number

Timo Brand, Henry Förster, Stephen Kobourov, Daniel Kohrt, Robin Schukrafft, Markus Wallinger, Johannes Zink

Graph drawing concerns the algorithmic visualization of graphs. A good drawing of a graph is easy to read and facilitates solving tasks on the graph. Several properties have been identified to occur in good drawings of graphs. Such properties include a low number of crossings, large angles between edges, short edges, and depicting symmetries. Many of these properties are explicitly measurable metrics. This lets us model a graph-drawing problem as a game where a single player iteratively moves vertices of a straight-line graph drawing to reduce edge crossings. We investigate whether reinforcement learning can discover effective strategies for playing this game. Our reinforcement-learning agent observes the local geometric and structural context of a vertex and selects a movement direction with the goal of reducing either the global or the local crossing number, that is, either the total number of crossings or the maximum number of crossings per edge. We compare the resulting strategies to existing methods and established crossing-minimization heuristics on standard benchmark graphs. While our approach does not out-compete state-of-the-art methods for minimizing the global crossing number, it is competitive and often superior for minimizing the local crossing number.

arXiv

CoCo: Code as CoT for Text-to-Image Preview and Rare Concept Generation

Haodong LI, Chunmei Qing, Huanyu Zhang, Dongzhi Jiang, Yihang Zou, Hongbo Peng, Dingming Li, Yuhong Dai, ZePeng Lin, Juanxi Tian, Yi Zhou, Siqi Dai, Jingwei Wu, Pheng-Ann Heng

Recent advancements in Unified Multimodal Models (UMMs) have significantly advanced text-to-image (T2I) generation, particularly through the integration of Chain-of-Thought (CoT) reasoning. However, existing CoT-based T2I methods largely rely on abstract natural-language planning, which lacks the precision required for complex spatial layouts, structured visual elements, and dense textual content. In this work, we propose CoCo (Code-as-CoT), a code-driven reasoning framework that represents the reasoning process as executable code, enabling explicit and verifiable intermediate planning for image generation. Given a text prompt, CoCo first generates executable code that specifies the structural layout of the scene. The code is then executed in a sandboxed environment to render a deterministic draft image. Subsequently, the model refines this draft through fine-grained image editing to produce the final high-fidelity result. To support this training paradigm, we construct CoCo-10K, a curated dataset containing structured draft-final image pairs designed to teach both structured draft construction and corrective visual refinement. Empirical evaluations on StructT2IBench, OneIG-Bench, and LongText-Bench show that CoCo achieves improvements of 68.83%, 54.8%, and 41.23%, respectively, over direct generation, while also outperforming other CoT-enhanced generation methods. These results demonstrate that executable code is an effective and reliable reasoning paradigm for precise, controllable, and structured text-to-image generation.

arXiv

Inference-Time Scaling of Diffusion Language Models via Trajectory Refinement

Meihua Dang, Jiaqi Han, Minkai Xu, Kai Xu, Akash Srivastava, Stefano Ermon

Discrete diffusion models have recently emerged as strong alternatives to autoregressive language models, matching their performance through large-scale training. However, inference-time control remains relatively underexplored. In this work, we study how to steer generation toward desired rewards without retraining the models. Prior methods typically resample or filter within a single denoising trajectory, optimizing rewards step-by-step without trajectory-level refinement. We introduce particle Gibbs sampling for diffusion language models (PG-DLM), an inference-time algorithm enabling trajectory-level refinement. PG-DLM constructs a Markov chain over full denoising trajectories and applies a conditional sequential Monte Carlo kernel to resample them. By doing so, PG-DLM introduces a new scaling axis, the number of refinement iterations, which is unavailable to prior methods. Increasing iterations remains effective even as gains from adding more parallel samples saturate. Furthermore, PG-DLM enables adaptive compute allocation by performing additional iterations only when needed, leading to further efficiency gains. We derive theoretical guarantees for convergence and variance bounds, and analyze trade-offs across different scaling axes. Empirically, PG-DLM outperforms prior methods across compute budgets on reward-guided generation tasks. On GSM8K, it achieves 90.07% accuracy with 2.9 particles on average and 94.47% accuracy with 24.8 particles on average.

arXiv

Agentic Software Issue Resolution with Large Language Models: A Survey

Zhonghao Jiang, David Lo, Zhongxin Liu

Software issue resolution aims to address real-world issues in software repositories based on natural language descriptions provided by users, and represents a key aspect of software maintenance. With the rapid development of large language models (LLMs) in reasoning and generation, LLM-based approaches have made significant progress in automated software issue resolution. However, resolving real-world software issues is inherently complex and requires long-horizon reasoning, iterative exploration, and feedback-driven decision-making, which demand agentic capabilities beyond conventional single-step approaches. Recently, LLM-based agentic systems have emerged as a promising research direction for software issue resolution, accompanied by rapid growth in the relevant literature. Advances in agentic software issue resolution can not only greatly improve the efficiency and quality of software maintenance, but also provide a realistic environment for evaluating the reasoning, planning, and execution capabilities of agentic systems, thereby bridging artificial intelligence and software engineering. This work presents a systematic survey of 242 recent studies at the forefront of research on LLM-based agentic software issue resolution. It outlines the general workflow of the task and establishes a taxonomy across three dimensions: benchmarks, techniques, and empirical studies. Furthermore, it highlights reinforcement learning as an increasingly important training paradigm for agentic systems in software engineering. Finally, it summarizes key challenges and outlines promising directions for future research.

arXiv

One Leak Away: How Pretrained Model Exposure Amplifies Jailbreak Risks in Finetuned LLMs

Yixin Tan, Zhe Yu, Rui Wen, Jun Sakuma

Finetuning pretrained large language models (LLMs) has become the standard paradigm for developing downstream applications. However, its security implications remain unclear, particularly regarding whether finetuned LLMs inherit jailbreak vulnerabilities from their pretrained sources. We investigate this question in a realistic pretrain-to-finetune threat model, where an attacker has full access to a released pretrained LLM but no access to its proprietary finetuned derivatives. Empirical analysis shows that adversarial prompts optimized on the pretrained model transfer most effectively to its finetuned variants, revealing inherited vulnerabilities from pretrained to finetuned LLMs. To further examine this inheritance, we conduct representation-level probing, which shows that transferable prompts are linearly separable within the pretrained hidden states, suggesting that transferability-relevant structure is already encoded in pretrained representations. Building on this insight, we propose the Probe-Guided Projection (PGP) attack, which steers optimization toward transferability-relevant directions. Experiments across multiple LLM families and diverse finetuned tasks confirm PGP's strong transfer success, underscoring the security risks inherent in the pretrain-to-finetune paradigm. Finally, we demonstrate that the same representation-level insights also enable a lightweight defense that mitigates pretrain-to-finetune jailbreak transfer while preserving downstream utility.

arXiv

Certified Interpolation Oversampling: Per-Instance Safety Guarantees for Imbalanced Learning

Pankaj Yadav, Vivek Vijay

Synthetic minority oversampling is typically designed and evaluated against a predictive objective, generating samples that improve downstream classification. This paper pursues a second objective by generating samples that carry a stated safety property, established for each instance by construction rather than assumed. We introduce Certified Interpolation Safe Oversampling (CISO), a three-phase interpolation framework built for this objective. A safety-guided distribution selects where synthesis occurs, a locality-and clearance-weighted distribution selects with whom each anchor interpolates, and a q-Gaussian placement density determines how far along the resulting segment each instance is placed. The framework provides three guarantees. Each synthetic instance seeded by a sufficiently safe anchor carries a certified distance from the majority class; a signed temperature parameter provably and monotonically shifts synthesis between boundary-seeking and interior-seeking regimes; and selection weights are strictly positive by construction, so no degenerate case arises. Certification is obtained alongside competitive predictive performance rather than in place of it. Under a protocol preregistered before evaluation, across 45 datasets, four classifiers, and eleven competing methods, CISO is statistically equivalent to SMOTE on precision-recall AUC, ranks second of eleven under gradient boosting, and completes every one of 11,460 fold-level evaluations without failure. A parameter sweep further reveals a continuous fidelity-safety trade-off that competing methods occupy only as isolated points.

arXiv

AegisShield: Democratizing Cyber Threat Modeling with Generative AI

Matthew Grofsky

The increasing sophistication of technology systems makes traditional threat modeling hard to scale, especially for small organizations with limited resources. This paper develops and evaluates AegisShield, a generative AI enhanced threat modeling tool that implements STRIDE and MITRE ATT&CK to automate threat generation and provide systematic assessments. By integrating real time threat intelligence from the National Vulnerability Database and AlienVault Open Threat Exchange, AegisShield produces streamlined and accessible threat descriptions. Our assessment of 243 threats from 15 case studies and over 8000 AI generated threats shows that AegisShield reduces complexity (p less than 0.001), yields outputs semantically aligned with expert developed threats (p less than 0.05), and achieves an 85.4 percent success rate in mapping threats to MITRE ATT&CK techniques (p less than 0.001). Automating and standardizing threat modeling helps under resourced organizations address risk earlier and supports wider adoption of secure by design practices.

arXiv

Convergence of Diffusion Models Under the Manifold Hypothesis in High-Dimensions

Iskander Azangulov, George Deligiannidis, Judith Rousseau

Denoising Diffusion Probabilistic Models (DDPM) are powerful state-of-the-art methods used to generate synthetic data from high-dimensional data distributions and are widely used for image, audio, and video generation as well as many more applications in science and beyond. The \textit{manifold hypothesis} states that high-dimensional data often lie on lower-dimensional manifolds within the ambient space, and is widely believed to hold in provided examples. While recent results have provided invaluable insight into how diffusion models adapt to the manifold hypothesis, they do not capture the great empirical success of these models, making this a very fruitful research direction. In this work, we study DDPMs under the manifold hypothesis and prove that they achieve rates independent of the ambient dimension in terms of score learning. In terms of sampling complexity, we obtain rates independent of the ambient dimension w.r.t.\ the Wasserstein distance. We do this by developing a new framework connecting diffusion models to the well-studied theory of extrema of Gaussian Processes.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.