English
Related papers

Related papers: Sink vs. diagonal patterns as mechanisms for atten…

200 papers

Transformers face quadratic complexity and memory issues with long sequences, prompting the adoption of linear attention mechanisms using fixed-size hidden states. However, linear models often suffer from limited recall performance, leading…

Computation and Language · Computer Science 2025-07-10 Dustin Wang , Rui-Jie Zhu , Steven Abreu , Yong Shan , Taylor Kergan , Yuqi Pan , Yuhong Chou , Zheng Li , Ge Zhang , Wenhao Huang , Jason Eshraghian

Recent work has made clear that the residual pathway is not mere optimization plumbing; it is part of the model's representational machinery. We agree, but argue that the cleanest way to organize this design space is through a two-axis view…

Machine Learning · Computer Science 2026-05-15 Yifan Zhang

Turbulent flows are fundamental in engineering and the environment, but their chaotic and three-dimensional (3-D) nature makes them computationally expensive to simulate. In this work, a dimensionality reduction technique is investigated to…

Fluid Dynamics · Physics 2020-12-16 Bernat Font

Self-supervised surround-view depth estimation enables dense, low-cost 3D perception with a 360{\deg} field of view from multiple minimally overlapping images. Yet, most existing methods suffer from depth estimates that are inconsistent…

Computer Vision and Pattern Recognition · Computer Science 2026-04-14 Samer Abualhanud , Christian Grannemann , Max Mehltretter

Dense retrieval conducts text retrieval in the embedding space and has shown many advantages compared to sparse retrieval. Existing dense retrievers optimize representations of queries and documents with contrastive training and map them to…

Information Retrieval · Computer Science 2021-07-19 Yizhi Li , Zhenghao Liu , Chenyan Xiong , Zhiyuan Liu

Various non-trivial spaces are becoming popular for embedding structured data such as graphs, texts, or images. Following spherical and hyperbolic spaces, more general product spaces have been proposed. However, searching for the best…

Machine Learning · Computer Science 2022-04-11 Kirill Shevkunov , Liudmila Prokhorenkova

Currently, learning better unsupervised sentence representations is the pursuit of many natural language processing communities. Lots of approaches based on pre-trained language models (PLMs) and contrastive learning have achieved promising…

Computation and Language · Computer Science 2023-05-11 Nuo Chen , Linjun Shou , Ming Gong , Jian Pei , Bowen Cao , Jianhui Chang , Daxin Jiang , Jia Li

We revisit a basic question in sequence modeling: is explicit self-attention actually necessary for strong performance and reasoning? We argue that standard multi-head attention is best seen as a form of tensor lifting: hidden vectors are…

Machine Learning · Computer Science 2025-12-23 Zhang Chong

We find that correct-to-incorrect sycophancy signals are most linearly separable within multi-head attention activations. Motivated by the linear representation hypothesis, we train linear probes across the residual stream, multilayer…

Computation and Language · Computer Science 2026-01-26 Rifo Genadi , Munachiso Nwadike , Nurdaulet Mukhituly , Hilal Alquabeh , Tatsuya Hiraoka , Kentaro Inui

Lattices are an efficient and effective method to encode ambiguity of upstream systems in natural language processing tasks, for example to compactly capture multiple speech recognition hypotheses, or to represent multiple linguistic…

Computation and Language · Computer Science 2019-06-05 Matthias Sperber , Graham Neubig , Ngoc-Quan Pham , Alex Waibel

Recent advances in transformer-based Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks. However, their quadratic computational complexity concerning sequence length remains a significant bottleneck…

Computation and Language · Computer Science 2025-06-05 Zichuan Fu , Wentao Song , Yejing Wang , Xian Wu , Yefeng Zheng , Yingying Zhang , Derong Xu , Xuetao Wei , Tong Xu , Xiangyu Zhao

Our study reveals new theoretical insights into over-smoothing and feature over-correlation in graph neural networks. Specifically, we demonstrate that with increased depth, node representations become dominated by a low-dimensional…

Machine Learning · Computer Science 2024-09-19 Andreas Roth , Thomas Liebig

Low rank tensor decompositions are a powerful tool for learning generative models, and uniqueness results give them a significant advantage over matrix decomposition methods. However, tensors pose significant algorithmic challenges and…

Data Structures and Algorithms · Computer Science 2014-01-21 Aditya Bhaskara , Moses Charikar , Ankur Moitra , Aravindan Vijayaraghavan

Attention mechanisms have been extensively employed in various applications, including time series modeling, owing to their capacity to capture intricate dependencies; however, their utility is often constrained by quadratic computational…

Machine Learning · Computer Science 2025-11-06 Mingtao Zhang , Guoli Yang , Zhanxing Zhu , Mengzhu Wang , Xiaoying Bai

Convolutional networks and vision transformers have different forms of pairwise interactions, pooling across layers and pooling at the end of the network. Does the latter really need to be different? As a by-product of pooling, vision…

Computer Vision and Pattern Recognition · Computer Science 2023-09-14 Bill Psomas , Ioannis Kakogeorgiou , Konstantinos Karantzalos , Yannis Avrithis

This paper investigates the symmetry properties of basins of attraction and their boundaries in equivariant dynamical systems. While the symmetry groups of compact attractors are well understood, the corresponding analysis for non-compact…

Chaotic Dynamics · Physics 2025-09-16 Xiao Xie

Blind-spot networks (BSN) have been prevalent neural architectures in self-supervised image denoising (SSID). However, most existing BSNs are conducted with convolution layers. Although transformers have shown the potential to overcome the…

Computer Vision and Pattern Recognition · Computer Science 2025-08-20 Junyi Li , Zhilu Zhang , Wangmeng Zuo

Transformers have achieved remarkable success in a wide range of natural language processing and computer vision applications. However, the representation capacity of a deep transformer model is degraded due to the over-smoothing issue in…

Computation and Language · Computer Science 2023-12-04 Tam Nguyen , Tan M. Nguyen , Richard G. Baraniuk

Recent advances in deep learning have relied heavily on the use of large Transformers due to their ability to learn at scale. However, the core building block of Transformers, the attention operator, exhibits quadratic cost in sequence…

Continual learning (CL) aims to train models that can sequentially learn new tasks without forgetting previous tasks' knowledge. Although previous works observed that pre-training can benefit CL, it remains unclear whether a pre-trained…

Machine Learning · Computer Science 2024-10-10 Xueying Bai , Yifan Sun , Niranjan Balasubramanian