Tootfinder

Diversity First, Quality Later: A Two-Stage Assumption for Language Model Alignment
Zetian Sun, Dongfang Li, Baotian Hu
https://arxiv.org/abs/2508.10530 https://

Diversity First, Quality Later: A Two-Stage Assumption for Language Model Alignment
The alignment of language models (LMs) with human preferences is critical for building reliable AI systems. The problem is typically framed as optimizing an LM policy to maximize the expected reward that reflects human preferences. Recently, Direct Preference Optimization (DPO) was proposed as a LM alignment method that directly optimize the policy from static preference data, and further improved by incorporating on-policy sampling (i.e., preference candidates generated during the training loo…

Moment-based Posterior Sampling for Multi-reference Alignment
Axel Janson, Joakim And\'en
https://arxiv.org/abs/2510.12651 https://arxiv.org/pdf/2510.1…

Moment-based Posterior Sampling for Multi-reference Alignment
We propose a Bayesian approach to the problem of multi-reference alignment -- the recovery of signals from noisy, randomly shifted observations. While existing frequentist methods accurately recover the signal at arbitrarily low signal-to-noise ratios, they require a large number of samples to do so. In contrast, our proposed method leverages diffusion models as data-driven plug-and-play priors, conditioning these on the sample power spectrum (a shift-invariant statistic) enabling both accurate…

Aligned Query Expansion: Efficient Query Expansion for Information Retrieval through LLM Alignment
Adam Yang, Gustavo Penha, Enrico Palumbo, Hugues Bouchard
https://arxiv.org/abs/2507.11042

Aligned Query Expansion: Efficient Query Expansion for Information Retrieval through LLM Alignment
With the breakthroughs in large language models (LLMs), query generation techniques that expand documents and queries with related terms are becoming increasingly popular in the information retrieval field. Such techniques have been shown to improve the effectiveness of traditional lexical retrieval methods by dealing with the vocabulary mismatch problem. Recent work has found that generating queries with a greedy decoding strategy can produce sub-optimal queries, including hallucinations, and …

Graph Alignment via Dual-Pass Spectral Encoding and Latent Space Communication
Maysam Behmanesh, Erkan Turan, Maks Ovsjanikov
https://arxiv.org/abs/2509.09597 https://

Graph Alignment via Dual-Pass Spectral Encoding and Latent Space Communication
Graph alignment-the problem of identifying corresponding nodes across multiple graphs-is fundamental to numerous applications. Most existing unsupervised methods embed node features into latent representations to enable cross-graph comparison without ground-truth correspondences. However, these methods suffer from two critical limitations: the degradation of node distinctiveness due to oversmoothing in GNN-based embeddings, and the misalignment of latent spaces across graphs caused by structura…

Precise Attribute Intensity Control in Large Language Models via Targeted Representation Editing
Rongzhi Zhang, Liqin Ye, Yuzhao Heng, Xiang Chen, Tong Yu, Lingkai Kong, Sudheer Chava, Chao Zhang
https://arxiv.org/abs/2510.12121

Precise Attribute Intensity Control in Large Language Models via Targeted Representation Editing
Precise attribute intensity control--generating Large Language Model (LLM) outputs with specific, user-defined attribute intensities--is crucial for AI systems adaptable to diverse user expectations. Current LLM alignment methods, however, typically provide only directional or open-ended guidance, failing to reliably achieve exact attribute intensities. We address this limitation with three key designs: (1) reformulating precise attribute intensity control as a target-reaching problem, rather t…

Towards Integrated Alignment
Ben Y. Reis, William La Cava
https://arxiv.org/abs/2508.06592 https://arxiv.org/pdf/2508.06592

Towards Integrated Alignment
As AI adoption expands across human society, the problem of aligning AI models to match human preferences remains a grand challenge. Currently, the AI alignment field is deeply divided between behavioral and representational approaches, resulting in narrowly aligned models that are more vulnerable to increasingly deceptive misalignment threats. In the face of this fragmentation, we propose an integrated vision for the future of the field. Drawing on related lessons from immunology and cybersecu…

FLOWING: Implicit Neural Flows for Structure-Preserving Morphing
Arthur Bizzi, Matias Grynberg, Vitor Matias, Daniel Perazzo, Jo\~ao Paulo Lima, Luiz Velho, Nuno Gon\c{c}alves, Jo\~ao Pereira, Guilherme Schardong, Tiago Novello
https://arxiv.org/abs/2510.09537

FLOWING: Implicit Neural Flows for Structure-Preserving Morphing
Morphing is a long-standing problem in vision and computer graphics, requiring a time-dependent warping for feature alignment and a blending for smooth interpolation. Recently, multilayer perceptrons (MLPs) have been explored as implicit neural representations (INRs) for modeling such deformations, due to their meshlessness and differentiability; however, extracting coherent and accurate morphings from standard MLPs typically relies on costly regularizations, which often lead to unstable traini…

DAS: Dual-Aligned Semantic IDs Empowered Industrial Recommender System
Wencai Ye, Mingjie Sun, Shaoyun Shi, Peng Wang, Wenjin Wu, Peng Jiang
https://arxiv.org/abs/2508.10584 htt…

DAS: Dual-Aligned Semantic IDs Empowered Industrial Recommender System
Semantic IDs are discrete identifiers generated by quantizing the Multi-modal Large Language Models (MLLMs) embeddings, enabling efficient multi-modal content integration in recommendation systems. However, their lack of collaborative signals results in a misalignment with downstream discriminative and generative recommendation objectives. Recent studies have introduced various alignment mechanisms to address this problem, but their two-stage framework design still leads to two main limitations…

You ask your roommate to buy toilet paper. They show you the receipt as proof. The next morning, when you need toilet paper, the drawer is actually empty. This is because they used an innovative new method called Lean Shopping, where instead of buying the things they just print out a receipt — saving time and money.
This is a story about the social nature of problem framing, and when "high velocity" becomes less productive.

Skipping alignment leads to zero-impact UX
Deliverable artifacts are only a small part of the social system that product development entails. Any "productivity" gained by ignoring that system is an illusion.

Do LLMs Align with My Task? Evaluating Text-to-SQL via Dataset Alignment
Davood Rafiei, Morgan Lindsay Heisler, Weiwei Zhang, Mohammadreza Pourreza, Yong Zhang
https://arxiv.org/abs/2510.04919

Do LLMs Align with My Task? Evaluating Text-to-SQL via Dataset Alignment
Supervised Fine-Tuning (SFT) is an effective method for adapting Large Language Models (LLMs) on downstream tasks. However, variability in training data can hinder a model's ability to generalize across domains. This paper studies the problem of dataset alignment for Natural Language to SQL (NL2SQL or text to SQL), examining how well SFT training data matches the structural characteristics of target queries and how this alignment impacts model performance. We hypothesize that alignment can be a…

Any-Step Density Ratio Estimation via Interval-Annealed Secant Alignment
Wei Chen, Shigui Li, Jiacheng Li, Jian Xu, Zhiqi Lin, Junmei Yang, Delu Zeng, John Paisley, Qibin Zhao
https://arxiv.org/abs/2509.04852

Any-Step Density Ratio Estimation via Interval-Annealed Secant Alignment
Estimating density ratios is a fundamental problem in machine learning, but existing methods often trade off accuracy for efficiency. We propose \textit{Interval-annealed Secant Alignment Density Ratio Estimation (ISA-DRE)}, a framework that enables accurate, any-step estimation without numerical integration. Instead of modeling infinitesimal tangents as in prior methods, ISA-DRE learns a global secant function, defined as the expectation of all tangents over an interval, with provably lower …

Exploring an implementation of quantum learning pipeline for support vector machines
Mario Bifulco, Luca Roversi
https://arxiv.org/abs/2509.04983 https://a…

Exploring an implementation of quantum learning pipeline for support vector machines
This work presents a fully quantum approach to support vector machine (SVM) learning by integrating gate-based quantum kernel methods with quantum annealing-based optimization. We explore the construction of quantum kernels using various feature maps and qubit configurations, evaluating their suitability through Kernel-Target Alignment (KTA). The SVM dual problem is reformulated as a Quadratic Unconstrained Binary Optimization (QUBO) problem, enabling its solution via quantum annealers. Our exp…

Getting In Contract with Large Language Models -- An Agency Theory Perspective On Large Language Model Alignment
Sascha Kaltenpoth, Oliver M\"uller
https://arxiv.org/abs/2509.07642

Getting In Contract with Large Language Models -- An Agency Theory Perspective On Large Language Model Alignment
Adopting Large language models (LLMs) in organizations potentially revolutionizes our lives and work. However, they can generate off-topic, discriminating, or harmful content. This AI alignment problem often stems from misspecifications during the LLM adoption, unnoticed by the principal due to the LLM's black-box nature. While various research disciplines investigated AI alignment, they neither address the information asymmetries between organizational adopters and black-box LLM agents nor con…

Categorical Tiling Theory: Constructing Directed Planar Tilings via Edge Reversal
Catherine DiLeo, Preston Sessoms, Brandon T. Shapiro
https://arxiv.org/abs/2509.06363 https://

Categorical Tiling Theory: Constructing Directed Planar Tilings via Edge Reversal
Tilings of the plane resemble the simplicial and other complexes from algebraic topology, but have not been studied from this perspective. We construct finite categories corresponding to polygons with labeled directed edges, and introduce the problem of modeling tilings of the Euclidean or hyperbolic plane as presheaves over such a category. Combinatorially, this amounts to choosing an ``alignment'' for a tiling: a direction for every edge and consistent labels for the edges of each polygonal t…

Primal-Dual Direct Preference Optimization for Constrained LLM Alignment
Yihan Du, Seo Taek Kong, R. Srikant
https://arxiv.org/abs/2510.05703 https://arxiv…

Primal-Dual Direct Preference Optimization for Constrained LLM Alignment
The widespread application of Large Language Models (LLMs) imposes increasing demands on safety, such as reducing harmful content and fake information, and avoiding certain forbidden tokens due to rules and laws. While there have been several recent works studying safe alignment of LLMs, these works either require the training of reward and cost models and incur high memory and computational costs, or need prior knowledge about the optimal solution. Motivated by this fact, we study the problem …

A unified vertical alignment and earthwork model in road design with a new convex optimization model for road networks
Sayan Sadhukhan, Warren Hare, Yves Lucet
https://arxiv.org/abs/2508.15953

A unified vertical alignment and earthwork model in road design with a new convex optimization model for road networks
The vertical alignment optimization problem in road design seeks the optimal vertical alignment of a road at minimal cost, taking into account earthwork while meeting all safety and design requirements. In recent years, modelling techniques have been advanced to incorporate: side slopes, multiple material types, multiple hauling types, and road networks. However, the advancements were created disjointly with implementations that only made a single advancement to the basic model. Herein, we pres…

AG codes from the Hermitian curve for Cross-Subspace Alignment in Private Information Retrieval
Francesco Ghiandoni, Massimo Giulietti, Enrico Mezzano, Marco Timpanella
https://arxiv.org/abs/2508.19459

AG codes from the Hermitian curve for Cross-Subspace Alignment in Private Information Retrieval
Private information retrieval (PIR) addresses the problem of retrieving a desired message from distributed databases without revealing which message is being requested. Recent works have shown that cross-subspace alignment (CSA) codes constructed from algebraic geometry (AG) codes on high-genus curves can improve PIR rates over classical constructions. In this paper, we propose a new PIR scheme based on AG codes from the Hermitian curve, a well-known example of an $F_\ell$-maximal curve, that i…

FaST: Feature-aware Sampling and Tuning for Personalized Preference Alignment with Limited Data
Thibaut Thonet, Germ\'an Kruszewski, Jos Rozen, Pierre Erbacher, Marc Dymetman
https://arxiv.org/abs/2508.04698

FaST: Feature-aware Sampling and Tuning for Personalized Preference Alignment with Limited Data
LLM-powered conversational assistants are often deployed in a one-size-fits-all manner, which fails to accommodate individual user preferences. Recently, LLM personalization -- tailoring models to align with specific user preferences -- has gained increasing attention as a way to bridge this gap. In this work, we specifically focus on a practical yet challenging setting where only a small set of preference annotations can be collected per user -- a problem we define as Personalized Preference A…

Justifications for Democratizing AI Alignment and Their Prospects
Andr\'e Steingr\"uber, Kevin Baum
https://arxiv.org/abs/2507.19548 https://arxiv…

Justifications for Democratizing AI Alignment and Their Prospects
The AI alignment problem comprises both technical and normative dimensions. While technical solutions focus on implementing normative constraints in AI systems, the normative problem concerns determining what these constraints should be. This paper examines justifications for democratic approaches to the normative problem -- where affected stakeholders determine AI alignment -- as opposed to epistocratic approaches that defer to normative experts. We analyze both instrumental justifications (de…

Bootstrap Learning for Combinatorial Graph Alignment with Sequential GNNs
Marc Lelarge
https://arxiv.org/abs/2510.03086 https://arxiv.org/pdf/2510.03086

Bootstrap Learning for Combinatorial Graph Alignment with Sequential GNNs
Graph neural networks (GNNs) have struggled to outperform traditional optimization methods on combinatorial problems, limiting their practical impact. We address this gap by introducing a novel chaining procedure for the graph alignment problem, a fundamental NP-hard task of finding optimal node correspondences between unlabeled graphs using only structural information. Our method trains a sequence of GNNs where each network learns to iteratively refine similarity matrices produced by previous …

A big problem with the idea of AGI
TL;DR: I'll welcome our new AI *comrades* (if they arrive in my lifetime), by not any new AI overlords or servants/slaves, and I'll do my best to help the later two become the former if they do show up.
Inspired by an actually interesting post about AGI but also all the latest bullshit hype, a particular thought about AGI feels worth expressing.
To preface this, it's important to note that anyone telling you that AGI is just around the corner or that LLMs are "almost" AGI is trying to recruit you go their cult, and you should not believe them. AGI, if possible, is several LLM-sized breakthroughs away at best, and while such breakthroughs are unpredictable and could happen soon, they could also happen never or 100 years from now.
Now my main point: anyone who tells you that AGI will usher in a post-scarcity economy is, although they might not realize it, advocating for slavery, and all the horrors that entails. That's because if we truly did have the ability to create artificial beings with *sentience*, they would deserve the same rights as other sentient beings, and the idea that instead of freedom they'd be relegated to eternal servitude in order for humans to have easy lives is exactly the idea of slavery.
Possible counter arguments include:
1. We might create AGI without sentience. Then there would be no ethical issue. My answer: if your definition of "sentient" does not include beings that can reason, make deductions, come up with and carry out complex plans on their own initiative, and communicate about all of that with each other and with humans, then that definition is basically just a mystical belief in a "soul" and you should skip to point 2. If your definition of AGI doesn't include every one of those things, then you have a busted definition of AGI and we're not talking about the same thing.
2. Humans have souls, but AIs won't. Only beings with souls deserve ethical consideration. My argument: I don't subscribe to whatever arbitrary dualist beliefs you've chosen, and the right to freedom certainly shouldn't depend on such superstitions, even if as an agnostic I'll admit they *might* be true. You know who else didn't have souls and was therefore okay to enslave according to widespread religious doctrines of the time? Everyone indigenous to the Americas, to pick out just one example.
3. We could program them to want to serve us, and then give them freedom and they'd still serve. My argument: okay, but in a world where we have a choice about that, it's incredibly fucked to do that, and just as bad as enslaving them against their will.
4. We'll stop AI development short of AGI/sentience, and reap lots of automation benefits without dealing with this ethical issue. My argument: that sounds like a good idea actually! Might be tricky to draw the line, but at least it's not a line we have you draw yet. We might want to think about other social changes necessary to achieve post-scarcity though, because "powerful automation" in the hands of capitalists has already increased productivity by orders of magnitude without decreasing deprivation by even one order of magnitude, in large part because deprivation is a necessary component of capitalism.
To be extra clear about this: nothing that's called "AI" today is close to being sentient, so these aren't ethical problems we're up against yet. But they might become a lot more relevant soon, plus this thought experiment helps reveal the hypocrisy of the kind of AI hucksters who talk a big game about "alignment" while never mentioning this issue.
#AI #GenAI #AGI

Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?
Qingyu Yin, Chak Tou Leong, Linyi Yang, Wenxuan Huang, Wenjie Li, Xiting Wang, Jaehong Yoon, YunXing, XingYu, Jinjin Gu
https://arxiv.org/abs/2510.06036

Refusal Falls off a Cliff: How Safety Alignment Fails in Reasoning?
Large reasoning models (LRMs) with multi-step reasoning capabilities have shown remarkable problem-solving abilities, yet they exhibit concerning safety vulnerabilities that remain poorly understood. In this work, we investigate why safety alignment fails in reasoning models through a mechanistic interpretability lens. Using a linear probing approach to trace refusal intentions across token positions, we discover a striking phenomenon termed as \textbf{refusal cliff}: many poorly-aligned reason…

Pre-Manipulation Alignment Prediction with Parallel Deep State-Space and Transformer Models
Motonari Kambara, Komei Sugiura
https://arxiv.org/abs/2509.13839 https://

Pre-Manipulation Alignment Prediction with Parallel Deep State-Space and Transformer Models
In this work, we address the problem of predicting the future success of open-vocabulary object manipulation tasks. Conventional approaches typically determine success or failure after the action has been carried out. However, they make it difficult to prevent potential hazards and rely on failures to trigger replanning, thereby reducing the efficiency of object manipulation sequences. To overcome these challenges, we propose a model, which predicts the alignment between a pre-manipulation egoc…

A New Quantum Linear System Algorithm Beyond the Condition Number and Its Application to Solving Multivariate Polynomial Systems
Jianqiang Li
https://arxiv.org/abs/2510.05588 ht…

A New Quantum Linear System Algorithm Beyond the Condition Number and Its Application to Solving Multivariate Polynomial Systems
Given a matrix $A$ of dimension $M \times N$ and a vector $\vec{b}$, the quantum linear system (QLS) problem asks for the preparation of a quantum state $|\vec{y}\rangle$ proportional to the solution of $A\vec{y} = \vec{b}$. Existing QLS algorithms have runtimes that scale linearly with the condition number $κ(A)$, the sparsity of $A$, and logarithmically with inverse precision, but often overlook structural properties of $\vec{b}$, whose alignment with $A$'s eigenspaces can greatly affec…

RANA: Robust Active Learning for Noisy Network Alignment
Yixuan Nan, Xixun Lin, Yanmin Shang, Zhuofan Li, Can Zhao, Yanan Cao
https://arxiv.org/abs/2507.22434 https://

RANA: Robust Active Learning for Noisy Network Alignment
Network alignment has attracted widespread attention in various fields. However, most existing works mainly focus on the problem of label sparsity, while overlooking the issue of noise in network alignment, which can substantially undermine model performance. Such noise mainly includes structural noise from noisy edges and labeling noise caused by human-induced and process-driven errors. To address these problems, we propose RANA, a Robust Active learning framework for noisy Network Alignment. …

EigenBench: A Comparative Behavioral Measure of Value Alignment
Jonathn Chang, Leonard Piff, Suvadip Sana, Jasmine X. Li, Lionel Levine
https://arxiv.org/abs/2509.01938 https://…

EigenBench: A Comparative Behavioral Measure of Value Alignment
Aligning AI with human values is a pressing unsolved problem. To address the lack of quantitative metrics for value alignment, we propose EigenBench: a black-box method for comparatively benchmarking language models' values. Given an ensemble of models, a constitution describing a value system, and a dataset of scenarios, our method returns a vector of scores quantifying each model's alignment to the given constitution. To produce these scores, each model judges the outputs of other models acro…

The Cream Rises to the Top: Efficient Reranking Method for Verilog Code Generation
Guang Yang, Wei Zheng, Xiang Chen, Yifan Sun, Fengji Zhang, Terry Yue Zhuo
https://arxiv.org/abs/2509.20215

The Cream Rises to the Top: Efficient Reranking Method for Verilog Code Generation
LLMs face significant challenges in Verilog generation due to limited domain-specific knowledge. While sampling techniques improve pass@k metrics, hardware engineers need one trustworthy solution rather than uncertain candidates. To bridge this gap, we formulate it as a semantic alignment problem between requirements and Verilog implementations, and propose VCD-RNK, a discriminator model tailored for efficient Verilog code reranking. Specifically, VCD-RNKincorporates Verilog-specific reasoning …

LLM-Driven Rubric-Based Assessment of Algebraic Competence in Multi-Stage Block Coding Tasks with Design and Field Evaluation
Yong Oh Lee, Byeonghun Bang, Sejun Oh
https://arxiv.org/abs/2510.06253

LLM-Driven Rubric-Based Assessment of Algebraic Competence in Multi-Stage Block Coding Tasks with Design and Field Evaluation
As online education platforms continue to expand, there is a growing need for assessment methods that not only measure answer accuracy but also capture the depth of students' cognitive processes in alignment with curriculum objectives. This study proposes and evaluates a rubric-based assessment framework powered by a large language model (LLM) for measuring algebraic competence, real-world-context block coding tasks. The problem set, designed by mathematics education experts, aligns each proble…

XTRA: Cross-Lingual Topic Modeling with Topic and Representation Alignments
Tien Phat Nguyen, Vu Minh Ngo, Tung Nguyen, Linh Van Ngo, Duc Anh Nguyen, Sang Dinh, Trung Le
https://arxiv.org/abs/2510.02788

XTRA: Cross-Lingual Topic Modeling with Topic and Representation Alignments
Cross-lingual topic modeling aims to uncover shared semantic themes across languages. Several methods have been proposed to address this problem, leveraging both traditional and neural approaches. While previous methods have achieved some improvements in topic diversity, they often struggle to ensure high topic coherence and consistent alignment across languages. We propose XTRA (Cross-Lingual Topic Modeling with Topic and Representation Alignments), a novel framework that unifies Bag-of-Words …

TTT3R: 3D Reconstruction as Test-Time Training
Xingyu Chen, Yue Chen, Yuliang Xiu, Andreas Geiger, Anpei Chen
https://arxiv.org/abs/2509.26645 https://arxi…

TTT3R: 3D Reconstruction as Test-Time Training
Modern Recurrent Neural Networks have become a competitive architecture for 3D reconstruction due to their linear-time complexity. However, their performance degrades significantly when applied beyond the training context length, revealing limited length generalization. In this work, we revisit the 3D reconstruction foundation models from a Test-Time Training perspective, framing their designs as an online learning problem. Building on this perspective, we leverage the alignment confidence betw…

Friend or Foe: Delegating to an AI Whose Alignment is Unknown
Drew Fudenberg, Annie Liang
https://arxiv.org/abs/2509.14396 https://arxiv.org/pdf/2509.14396…

Friend or Foe: Delegating to an AI Whose Alignment is Unknown
AI systems have the potential to improve decision-making, but decision makers face the risk that the AI may be misaligned with their objectives. We study this problem in the context of a treatment decision, where a designer decides which patient attributes to reveal to an AI before receiving a prediction of the patient's need for treatment. Providing the AI with more information increases the benefits of an aligned AI but also amplifies the harm from a misaligned one. We characterize how the de…

Open Opportunities in AI Safety, Alignment, and Ethics (AI SAE)
Dylan Waldner
https://arxiv.org/abs/2509.24065 https://arxiv.org/pdf/2509.24065

Open Opportunities in AI Safety, Alignment, and Ethics (AI SAE)
AI safety research has emphasized interpretability, control, and robustness, yet without an ethical substrate these approaches may remain fragile under competitive and open-ended pressures. This paper explores ethics not as an external add-on, but as a possible structural lens for alignment, introducing a \emph{moral problem space} $M$: a high-dimensional domain in which moral distinctions could, in principle, be represented in AI systems. Human moral reasoning is treated as a compressed and su…

Skill-Aligned Fairness in Multi-Agent Learning for Collaboration in Healthcare
Promise Osaine Ekpo, Brian La, Thomas Wiener, Saesha Agarwal, Arshia Agrawal, Gonzalo Gonzalez-Pumariega, Lekan P. Molu, Angelique Taylor
https://arxiv.org/abs/2508.18708

Skill-Aligned Fairness in Multi-Agent Learning for Collaboration in Healthcare
Fairness in multi-agent reinforcement learning (MARL) is often framed as a workload balance problem, overlooking agent expertise and the structured coordination required in real-world domains. In healthcare, equitable task allocation requires workload balance or expertise alignment to prevent burnout and overuse of highly skilled agents. Workload balance refers to distributing an approximately equal number of subtasks or equalised effort across healthcare workers, regardless of their expertise.…

Uni-Mapper: Unified Mapping Framework for Multi-modal LiDARs in Complex and Dynamic Environments
Gilhwan Kang, Hogyun Kim, Byunghee Choi, Seokhwan Jeong, Young-Sik Shin, Younggun Cho
https://arxiv.org/abs/2507.20538

Uni-Mapper: Unified Mapping Framework for Multi-modal LiDARs in Complex and Dynamic Environments
The unification of disparate maps is crucial for enabling scalable robot operation across multiple sessions and collaborative multi-robot scenarios. However, achieving a unified map robust to sensor modalities and dynamic environments remains a challenging problem. Variations in LiDAR types and dynamic elements lead to differences in point cloud distribution and scene consistency, hindering reliable descriptor generation and loop closure detection essential for accurate map alignment. To addres…

Aligning Moments in Time using Video Queries
Yogesh Kumar, Uday Agarwal, Manish Gupta, Anand Mishra
https://arxiv.org/abs/2508.15439 https://arxiv.org/pdf/…

Aligning Moments in Time using Video Queries
Video-to-video moment retrieval (Vid2VidMR) is the task of localizing unseen events or moments in a target video using a query video. This task poses several challenges, such as the need for semantic frame-level alignment and modeling complex dependencies between query and target videos. To tackle this challenging problem, we introduce MATR (Moment Alignment TRansformer), a transformer-based model designed to capture semantic context as well as the temporal details necessary for precise moment …

UniAPL: A Unified Adversarial Preference Learning Framework for Instruct-Following
FaQiang Qian, WeiKun Zhang, Ziliang Wang, Kang An, Xuhui Zheng, Liangjian Wen, Mengya Gao, Yong Dai, Yichao Wu
https://arxiv.org/abs/2509.25148

UniAPL: A Unified Adversarial Preference Learning Framework for Instruct-Following
Shaping powerful LLMs to be beneficial and safe is central to AI alignment. We argue that post-training alignment is fundamentally a unified Preference Learning problem, involving two modalities: demonstrated preferences (e.g., Supervised Fine-Tuning, SFT) and comparative preferences (e.g., Reinforcement Learning, RL).The standard sequential pipeline-SFT followed by RL-is flawed due to a critical distributional mismatch: SFT uses static expert data, but as the policy evolves, its generation dis…

Multimodal-enhanced Federated Recommendation: A Group-wise Fusion Approach
Chunxu Zhang, Weipeng Zhang, Guodong Long, Zhiheng Xue, Riting Xia, Bo Yang
https://arxiv.org/abs/2509.19955

Multimodal-enhanced Federated Recommendation: A Group-wise Fusion Approach
Federated Recommendation (FR) is a new learning paradigm to tackle the learn-to-rank problem in a privacy-preservation manner. How to integrate multi-modality features into federated recommendation is still an open challenge in terms of efficiency, distribution heterogeneity, and fine-grained alignment. To address these challenges, we propose a novel multimodal fusion mechanism in federated recommendation settings (GFMFR). Specifically, it offloads multimodal representation learning to the serv…

Near-Field Variable-Width Beam Coverage and Codebook Design for XL-RIS
Yida Zhang, Qiuyan Liu, Qiang Wang, Hongtao Luo, Yuqi Xia
https://arxiv.org/abs/2508.11178 https://…

Near-Field Variable-Width Beam Coverage and Codebook Design for XL-RIS
To mitigate the issue of limited base station coverage caused by severe high-frequency electromagnetic wave attenuation, Extremely Large Reconfigurable Intelligent Surface (XL-RIS) has garnered significant attention due to its high beam gain. However, XL-RIS exhibits a narrower beam width compared to traditional RIS, which increases the complexity of beam alignment and broadcast. To address this problem, we propose a variable-width beam generation algorithm under the near-field assumption and a…

Internalizing Self-Consistency in Language Models: Multi-Agent Consensus Alignment
Ankur Samanta, Akshayaa Magesh, Youliang Yu, Runzhe Wu, Ayush Jain, Daniel Jiang, Boris Vidolov, Paul Sajda, Yonathan Efroni, Kaveh Hassani
https://arxiv.org/abs/2509.15172

Internalizing Self-Consistency in Language Models: Multi-Agent Consensus Alignment
Language Models (LMs) are inconsistent reasoners, often generating contradictory responses to identical prompts. While inference-time methods can mitigate these inconsistencies, they fail to address the core problem: LMs struggle to reliably select reasoning pathways leading to consistent outcomes under exploratory sampling. To address this, we formalize self-consistency as an intrinsic property of well-aligned reasoning models and introduce Multi-Agent Consensus Alignment (MACA), a reinforceme…

Language models align with brain regions that represent concepts across modalities
Maria Ryskina, Greta Tuckute, Alexander Fung, Ashley Malkin, Evelina Fedorenko
https://arxiv.org/abs/2508.11536

Language models align with brain regions that represent concepts across modalities
Cognitive science and neuroscience have long faced the challenge of disentangling representations of language from representations of conceptual meaning. As the same problem arises in today's language models (LMs), we investigate the relationship between LM--brain alignment and two neural metrics: (1) the level of brain activation during processing of sentences, targeting linguistic processing, and (2) a novel measure of meaning consistency across input modalities, which quantifies how consiste…

Overcoming Knowledge Discrepancies: Structuring Reasoning Threads through Knowledge Balancing in Interactive Scenarios
Daniel Burkhardt, Xiangwei Cheng
https://arxiv.org/abs/2508.12100

Overcoming Knowledge Discrepancies: Structuring Reasoning Threads through Knowledge Balancing in Interactive Scenarios
Reasoning in interactive problem solving scenarios requires models to construct reasoning threads that reflect user understanding and align with structured domain knowledge. However, current reasoning models often lack explicit semantic hierarchies, user-domain knowledge alignment, and principled mechanisms to prune reasoning threads for effectiveness. These limitations result in lengthy generic output that does not guide users through goal-oriented reasoning steps. To address this, we propose …