SEARCH · Search arXiv

Results for “transformer architecture”

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 415 records · Page 23Linked to original sources

arXiv2024

A Hyper-Transformer model for Controllable Pareto Front Learning with Split Feasibility Constraints

Controllable Pareto front learning (CPFL) approximates the Pareto solution set and then locates a Pareto optimal solution with respect to a given reference vector. However, decision-maker objectives were limited to a constraint region in practice, so instead of training on the entire decision space, we only trained on the constraint region. Controllable Pareto front learning with Split Feasibility Constraints (SFC) is a way to find the best Pareto solutions to a split multi-objective optimization problem that meets certain constraints. In the previous study, CPFL used a Hypernetwork model comprising multi-layer perceptron (Hyper-MLP) blocks. With the substantial advancement of transformer architecture in deep learning, transformers can outperform other architectures in various tasks. Therefore, we have developed a hyper-transformer (Hyper-Trans) model for CPFL with SFC. We use the theory of universal approximation for the sequence-to-sequence function to show that the Hyper-Trans model makes MED errors smaller in computational experiments than the Hyper-MLP model.

cs.LG↗

arXiv2023

Evaluating self-attention interpretability through human-grounded experimental protocol

Attention mechanisms have played a crucial role in the development of complex architectures such as Transformers in natural language processing. However, Transformers remain hard to interpret and are considered as black-boxes. This paper aims to assess how attention coefficients from Transformers can help in providing interpretability. A new attention-based interpretability method called CLaSsification-Attention (CLS-A) is proposed. CLS-A computes an interpretability score for each word based on the attention coefficient distribution related to the part specific to the classification task within the Transformer architecture. A human-grounded experiment is conducted to evaluate and compare CLS-A to other interpretability methods. The experimental protocol relies on the capacity of an interpretability method to provide explanation in line with human reasoning. Experiment design includes measuring reaction times and correct response rates by human subjects. CLS-A performs comparably to usual interpretability methods regarding average participant reaction time and accuracy. The lower computational cost of CLS-A compared to other interpretability methods and its availability by design within the classifier make it particularly interesting. Data analysis also highlights the link between the probability score of a classifier prediction and adequate explanations. Finally, our work confirms the relevancy of the use of CLS-A and shows to which extent self-attention contains rich information to explain Transformer classifiers.

cs.LG↗

arXiv2025

MTDP: A Modulated Transformer based Diffusion Policy Model

Recent research on robot manipulation based on Behavior Cloning (BC) has made significant progress. By combining diffusion models with BC, diffusion policiy has been proposed, enabling robots to quickly learn manipulation tasks with high success rates. However, integrating diffusion policy with high-capacity Transformer presents challenges, traditional Transformer architectures struggle to effectively integrate guiding conditions, resulting in poor performance in manipulation tasks when using Transformer-based models. In this paper, we investigate key architectural designs of Transformers and improve the traditional Transformer architecture by proposing the Modulated Transformer Diffusion Policy (MTDP) model for diffusion policy. The core of this model is the Modulated Attention module we proposed, which more effectively integrates the guiding conditions with the main input, improving the generative model's output quality and, consequently, increasing the robot's task success rate. In six experimental tasks, MTDP outperformed existing Transformer model architectures, particularly in the Toolhang experiment, where the success rate increased by 12\%. To verify the generality of Modulated Attention, we applied it to the UNet architecture to construct Modulated UNet Diffusion Policy model (MUDP), which also achieved higher success rates than existing UNet architectures across all six experiments. The Diffusion Policy uses Denoising Diffusion Probabilistic Models (DDPM) as the diffusion model. Building on this, we also explored Denoising Diffusion Implicit Models (DDIM) as the diffusion model, constructing the MTDP-I and MUDP-I model, which nearly doubled the generation speed while maintaining performance.

cs.RO↗

arXiv2020

Towards Fully 8-bit Integer Inference for the Transformer Model

8-bit integer inference, as a promising direction in reducing both the latency and storage of deep neural networks, has made great progress recently. On the other hand, previous systems still rely on 32-bit floating point for certain functions in complex models (e.g., Softmax in Transformer), and make heavy use of quantization and de-quantization. In this work, we show that after a principled modification on the Transformer architecture, dubbed Integer Transformer, an (almost) fully 8-bit integer inference algorithm Scale Propagation could be derived. De-quantization is adopted when necessary, which makes the network more efficient. Our experiments on WMT16 En<->Ro, WMT14 En<->De and En->Fr translation tasks as well as the WikiText-103 language modelling task show that the fully 8-bit Transformer system achieves comparable performance with the floating point baseline but requires nearly 4x less memory footprint.

cs.CL↗

arXiv2022

The Quarks of Attention

Attention plays a fundamental role in both natural and artificial intelligence systems. In deep learning, attention-based neural architectures, such as transformer architectures, are widely used to tackle problems in natural language processing and beyond. Here we investigate the fundamental building blocks of attention and their computational properties. Within the standard model of deep learning, we classify all possible fundamental building blocks of attention in terms of their source, target, and computational mechanism. We identify and study three most important mechanisms: additive activation attention, multiplicative output attention (output gating), and multiplicative synaptic attention (synaptic gating). The gating mechanisms correspond to multiplicative extensions of the standard model and are used across all current attention-based deep learning architectures. We study their functional properties and estimate the capacity of several attentional building blocks in the case of linear and polynomial threshold gates. Surprisingly, additive activation attention plays a central role in the proofs of the lower bounds. Attention mechanisms reduce the depth of certain basic circuits and leverage the power of quadratic activations without incurring their full cost.

cs.LG↗

arXiv2022

Graph Flow: Cross-layer Graph Flow Distillation for Dual Efficient Medical Image Segmentation

With the development of deep convolutional neural networks, medical image segmentation has achieved a series of breakthroughs in recent years. However, the high-performance convolutional neural networks always mean numerous parameters and high computation costs, which will hinder the applications in clinical scenarios. Meanwhile, the scarceness of large-scale annotated medical image datasets further impedes the application of high-performance networks. To tackle these problems, we propose Graph Flow, a comprehensive knowledge distillation framework, for both network-efficiency and annotation-efficiency medical image segmentation. Specifically, our core Graph Flow Distillation transfer the essence of cross-layer variations from a well-trained cumbersome teacher network to a non-trained compact student network. In addition, an unsupervised Paraphraser Module is integrated to purify the knowledge of the teacher network, which is also beneficial for the stabilization of training procedure. Furthermore, we build a unified distillation framework by integrating the adversarial distillation and the vanilla logits distillation, which can further refine the final predictions of the compact network. With different teacher networks (conventional convolutional architecture or prevalent transformer architecture) and student networks, we conduct extensive experiments on four medical image datasets with different modalities (Gastric Cancer, Synapse, BUSI, and CVC-ClinicDB).We demonstrate the prominent ability of our method which achieves competitive performance on these datasets. Moreover, we demonstrate the effectiveness of our Graph Flow through a novel semi-supervised paradigm for dual efficient medical image segmentation. Our code will be available at Graph Flow.

cs.CV↗

arXiv2022

Fooling Explanations in Text Classifiers

State-of-the-art text classification models are becoming increasingly reliant on deep neural networks (DNNs). Due to their black-box nature, faithful and robust explanation methods need to accompany classifiers for deployment in real-life scenarios. However, it has been shown in vision applications that explanation methods are susceptible to local, imperceptible perturbations that can significantly alter the explanations without changing the predicted classes. We show here that the existence of such perturbations extends to text classifiers as well. Specifically, we introduceTextExplanationFooler (TEF), a novel explanation attack algorithm that alters text input samples imperceptibly so that the outcome of widely-used explanation methods changes considerably while leaving classifier predictions unchanged. We evaluate the performance of the attribution robustness estimation performance in TEF on five sequence classification datasets, utilizing three DNN architectures and three transformer architectures for each dataset. TEF can significantly decrease the correlation between unchanged and perturbed input attributions, which shows that all models and explanation methods are susceptible to TEF perturbations. Moreover, we evaluate how the perturbations transfer to other model architectures and attribution methods, and show that TEF perturbations are also effective in scenarios where the target model and explanation method are unknown. Finally, we introduce a semi-universal attack that is able to compute fast, computationally light perturbations with no knowledge of the attacked classifier nor explanation method. Overall, our work shows that explanations in text classifiers are very fragile and users need to carefully address their robustness before relying on them in critical applications.

cs.LG↗

arXiv2023

KINLP at SemEval-2023 Task 12: Kinyarwanda Tweet Sentiment Analysis

This paper describes the system entered by the author to the SemEval-2023 Task 12: Sentiment analysis for African languages. The system focuses on the Kinyarwanda language and uses a language-specific model. Kinyarwanda morphology is modeled in a two tier transformer architecture and the transformer model is pre-trained on a large text corpus using multi-task masked morphology prediction. The model is deployed on an experimental platform that allows users to experiment with the pre-trained language model fine-tuning without the need to write machine learning code. Our final submission to the shared task achieves second ranking out of 34 teams in the competition, achieving 72.50% weighted F1 score. Our analysis of the evaluation results highlights challenges in achieving high accuracy on the task and identifies areas for improvement.

cs.CL↗

arXiv2023

Enhancing Short-Term Wind Speed Forecasting using Graph Attention and Frequency-Enhanced Mechanisms

The safe and stable operation of power systems is greatly challenged by the high variability and randomness of wind power in large-scale wind-power-integrated grids. Wind power forecasting is an effective solution to tackle this issue, with wind speed forecasting being an essential aspect. In this paper, a Graph-attentive Frequency-enhanced Spatial-Temporal Wind Speed Forecasting model based on graph attention and frequency-enhanced mechanisms, i.e., GFST-WSF, is proposed to improve the accuracy of short-term wind speed forecasting. The GFST-WSF comprises a Transformer architecture for temporal feature extraction and a Graph Attention Network (GAT) for spatial feature extraction. The GAT is specifically designed to capture the complex spatial dependencies among wind speed stations to effectively aggregate information from neighboring nodes in the graph, thus enhancing the spatial representation of the data. To model the time lag in wind speed correlation between adjacent wind farms caused by geographical factors, a dynamic complex adjacency matrix is formulated and utilized by the GAT. Benefiting from the effective spatio-temporal feature extraction and the deep architecture of the Transformer, the GFST-WSF outperforms other baselines in wind speed forecasting for the 6-24 hours ahead forecast horizon in case studies.

cs.LG↗

arXiv2023

LEST: Large-scale LiDAR Semantic Segmentation with Transformer

Large-scale LiDAR-based point cloud semantic segmentation is a critical task in autonomous driving perception. Almost all of the previous state-of-the-art LiDAR semantic segmentation methods are variants of sparse 3D convolution. Although the Transformer architecture is becoming popular in the field of natural language processing and 2D computer vision, its application to large-scale point cloud semantic segmentation is still limited. In this paper, we propose a LiDAR sEmantic Segmentation architecture with pure Transformer, LEST. LEST comprises two novel components: a Space Filling Curve (SFC) Grouping strategy and a Distance-based Cosine Linear Transformer, DISCO. On the public nuScenes semantic segmentation validation set and SemanticKITTI test set, our model outperforms all the other state-of-the-art methods.

cs.CV↗

arXiv2024

YotoR-You Only Transform One Representation

This paper introduces YotoR (You Only Transform One Representation), a novel deep learning model for object detection that combines Swin Transformers and YoloR architectures. Transformers, a revolutionary technology in natural language processing, have also significantly impacted computer vision, offering the potential to enhance accuracy and computational efficiency. YotoR combines the robust Swin Transformer backbone with the YoloR neck and head. In our experiments, YotoR models TP5 and BP4 consistently outperform YoloR P6 and Swin Transformers in various evaluations, delivering improved object detection performance and faster inference speeds than Swin Transformer models. These results highlight the potential for further model combinations and improvements in real-time object detection with Transformers. The paper concludes by emphasizing the broader implications of YotoR, including its potential to enhance transformer-based models for image-related tasks.

cs.CV↗

arXiv2024

Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models and Time-Dependent Layer Normalization

This paper presents innovative enhancements to diffusion models by integrating a novel multi-resolution network and time-dependent layer normalization. Diffusion models have gained prominence for their effectiveness in high-fidelity image generation. While conventional approaches rely on convolutional U-Net architectures, recent Transformer-based designs have demonstrated superior performance and scalability. However, Transformer architectures, which tokenize input data (via "patchification"), face a trade-off between visual fidelity and computational complexity due to the quadratic nature of self-attention operations concerning token length. While larger patch sizes enable attention computation efficiency, they struggle to capture fine-grained visual details, leading to image distortions. To address this challenge, we propose augmenting the Diffusion model with the Multi-Resolution network (DiMR), a framework that refines features across multiple resolutions, progressively enhancing detail from low to high resolution. Additionally, we introduce Time-Dependent Layer Normalization (TD-LN), a parameter-efficient approach that incorporates time-dependent parameters into layer normalization to inject time information and achieve superior performance. Our method's efficacy is demonstrated on the class-conditional ImageNet generation benchmark, where DiMR-XL variants outperform prior diffusion models, setting new state-of-the-art FID scores of 1.70 on ImageNet 256 x 256 and 2.89 on ImageNet 512 x 512. Project page: https://qihao067.github.io/projects/DiMR

cs.CV↗

arXiv2024

Continuous-Time Linear Positional Embedding for Irregular Time Series Forecasting

Irregularly sampled time series forecasting, characterized by non-uniform intervals, is prevalent in practical applications. However, previous research have been focused on regular time series forecasting, typically relying on transformer architectures. To extend transformers to handle irregular time series, we tackle the positional embedding which represents the temporal information of the data. We propose CTLPE, a method learning a continuous linear function for encoding temporal information. The two challenges of irregular time series, inconsistent observation patterns and irregular time gaps, are solved by learning a continuous-time function and concise representation of position. Additionally, the linear continuous function is empirically shown superior to other continuous functions by learning a neural controlled differential equation-based positional embedding, and theoretically supported with properties of ideal positional embedding. CTLPE outperforms existing techniques across various irregularly-sampled time series datasets, showcasing its enhanced efficacy.

cs.LG↗

arXiv2024

TGTOD: A Global Temporal Graph Transformer for Outlier Detection at Scale

While Transformers have revolutionized machine learning on various data, existing Transformers for temporal graphs face limitations in (1) restricted receptive fields, (2) overhead of subgraph extraction, and (3) suboptimal generalization capability beyond link prediction. In this paper, we rethink temporal graph Transformers and propose TGTOD, a novel end-to-end Temporal Graph Transformer for Outlier Detection. TGTOD employs global attention to model both structural and temporal dependencies within temporal graphs. To tackle scalability, our approach divides large temporal graphs into spatiotemporal patches, which are then processed by a hierarchical Transformer architecture comprising Patch Transformer, Cluster Transformer, and Temporal Transformer. We evaluate TGTOD on three public datasets under two settings, comparing with a wide range of baselines. Our experimental results demonstrate the effectiveness of TGTOD, achieving AP improvement of 61% on Elliptic. Furthermore, our efficiency evaluation shows that TGTOD reduces training time by 44x compared to existing Transformers for temporal graphs. To foster reproducibility, we make our implementation publicly available at https://github.com/kayzliu/tgtod.

cs.LG↗

arXiv2024

Global Average Feature Augmentation for Robust Semantic Segmentation with Transformers

Robustness to out-of-distribution data is crucial for deploying modern neural networks. Recently, Vision Transformers, such as SegFormer for semantic segmentation, have shown impressive robustness to visual corruptions like blur or noise affecting the acquisition device. In this paper, we propose Channel Wise Feature Augmentation (CWFA), a simple yet efficient feature augmentation technique to improve the robustness of Vision Transformers for semantic segmentation. CWFA applies a globally estimated perturbation per encoder with minimal compute overhead during training. Extensive evaluations on Cityscapes and ADE20K, with three state-of-the-art Vision Transformer architectures : SegFormer, Swin Transformer, and Twins demonstrate that CWFA-enhanced models significantly improve robustness without affecting clean data performance. For instance, on Cityscapes, a CWFA-augmented SegFormer-B1 model yields up to 27.7% mIoU robustness gain on impulse noise compared to the non-augmented SegFormer-B1. Furthermore, CWFA-augmented SegFormer-B5 achieves a new state-of-the-art 84.3% retention rate, a 0.7% improvement over the recently published FAN+STL.

cs.CV↗

arXiv2025

Geometric Feature Prompting of Image Segmentation Models

Advances in machine learning, especially the introduction of transformer architectures and vision transformers, have led to the development of highly capable computer vision foundation models. The segment anything model (known colloquially as SAM and more recently SAM 2), is a highly capable foundation model for segmentation of natural images and has been further applied to medical and scientific image segmentation tasks. SAM relies on prompts -- points or regions of interest in an image -- to generate associated segmentations. In this manuscript we propose the use of a geometrically motivated prompt generator to produce prompt points that are colocated with particular features of interest. Focused prompting enables the automatic generation of sensitive and specific segmentations in a scientific image analysis task using SAM with relatively few point prompts. The image analysis task examined is the segmentation of plant roots in rhizotron or minirhizotron images, which has historically been a difficult task to automate. Hand annotation of rhizotron images is laborious and often subjective; SAM, initialized with GeomPrompt local ridge prompts has the potential to dramatically improve rhizotron image processing. The authors have concurrently released an open source software suite called geomprompt https://pypi.org/project/geomprompt/ that can produce point prompts in a format that enables direct integration with the segment-anything package.

cs.CV↗

arXiv2026

Transformers Provably Learn Algorithmic Solutions for Graph Connectivity, But Only with the Right Data

Transformers often fail to learn generalizable algorithms, instead relying on brittle heuristics. Using graph connectivity as a testbed, we explain this phenomenon both theoretically and empirically. We consider a simplified Transformer architecture, the Disentangled Transformer, and prove that an $L$-layer model can compute connectivity in graphs with diameters up to $3^L$, implementing an algorithm equivalent to computing powers of the adjacency matrix. By analyzing training dynamics, we prove that whether the model learns this strategy hinges on whether most training instances are within this model capacity. Within-capacity graphs (diameter $\leq 3^L$) drive the learning of the algorithmic solution while beyond-capacity graphs drive the learning of a simple heuristic based on node degrees. Finally, we empirically show that restricting training data to stay within a model's capacity makes both standard and Disentangled Transformers learn the exact algorithm.

cs.LG↗

arXiv2026

WUTDet: A 100K-Scale Ship Detection Dataset and Benchmarks with Dense Small Objects

Ship detection for navigation is a fundamental perception task in intelligent waterway transportation systems. However, existing public ship detection datasets remain limited in terms of scale, the proportion of small-object instances, and scene diversity, which hinders the systematic evaluation and generalization study of detection algorithms in complex maritime environments. To this end, we construct WUTDet, a large-scale ship detection dataset. WUTDet contains 100,576 images and 381,378 annotated ship instances, covering diverse operational scenarios such as ports, anchorages, navigation, and berthing, as well as various imaging conditions including fog, glare, low-lightness, and rain, thereby exhibiting substantial diversity and challenge. Based on WUTDet, we systematically evaluate 20 baseline models from three mainstream detection architectures, namely CNN, Transformer, and Mamba. Experimental results show that the Transformer architecture achieves superior overall detection accuracy (AP) and small-object detection performance (APs), demonstrating stronger adaptability to complex maritime scenes; the CNN architecture maintains an advantage in inference efficiency, making it more suitable for real-time applications; and the Mamba architecture achieves a favorable balance between detection accuracy and computational efficiency. Furthermore, we construct a unified cross-dataset test set, Ship-GEN, to evaluate model generalization. Results on Ship-GEN show that models trained on WUTDet exhibit stronger generalization under different data distributions. These findings demonstrate that WUTDet provides effective data support for the research, evaluation, and generalization analysis of ship detection algorithms in complex maritime scenarios. The dataset is publicly available at: https://github.com/MAPGroup/WUTDet.

cs.CV↗