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 1,153 records · Page 64Linked to original sources

arXiv2023

Scalable Diffusion Models with Transformers

We explore a new class of diffusion models based on the transformer architecture. We train latent diffusion models of images, replacing the commonly-used U-Net backbone with a transformer that operates on latent patches. We analyze the scalability of our Diffusion Transformers (DiTs) through the lens of forward pass complexity as measured by Gflops. We find that DiTs with higher Gflops -- through increased transformer depth/width or increased number of input tokens -- consistently have lower FID. In addition to possessing good scalability properties, our largest DiT-XL/2 models outperform all prior diffusion models on the class-conditional ImageNet 512x512 and 256x256 benchmarks, achieving a state-of-the-art FID of 2.27 on the latter.

cs.CV↗

arXiv2023

A transformer-based approach to video frame-level prediction in Affective Behaviour Analysis In-the-wild

In recent years, transformer architecture has been a dominating paradigm in many applications, including affective computing. In this report, we propose our transformer-based model to handle Emotion Classification Task in the 5th Affective Behavior Analysis In-the-wild Competition. By leveraging the attentive model and the synthetic dataset, we attain a score of 0.4775 on the validation set of Aff-Wild2, the dataset provided by the organizer.

cs.CV↗

arXiv2023

ConvFormer: Parameter Reduction in Transformer Models for 3D Human Pose Estimation by Leveraging Dynamic Multi-Headed Convolutional Attention

Recently, fully-transformer architectures have replaced the defacto convolutional architecture for the 3D human pose estimation task. In this paper we propose \textbf{\textit{ConvFormer}}, a novel convolutional transformer that leverages a new \textbf{\textit{dynamic multi-headed convolutional self-attention}} mechanism for monocular 3D human pose estimation. We designed a spatial and temporal convolutional transformer to comprehensively model human joint relations within individual frames and globally across the motion sequence. Moreover, we introduce a novel notion of \textbf{\textit{temporal joints profile}} for our temporal ConvFormer that fuses complete temporal information immediately for a local neighborhood of joint features. We have quantitatively and qualitatively validated our method on three common benchmark datasets: Human3.6M, MPI-INF-3DHP, and HumanEva. Extensive experiments have been conducted to identify the optimal hyper-parameter set. These experiments demonstrated that we achieved a \textbf{significant parameter reduction relative to prior transformer models} while attaining State-of-the-Art (SOTA) or near SOTA on all three datasets. Additionally, we achieved SOTA for Protocol III on H36M for both GT and CPN detection inputs. Finally, we obtained SOTA on all three metrics for the MPI-INF-3DHP dataset and for all three subjects on HumanEva under Protocol II.

cs.CV↗

arXiv2023

Transformers in Reinforcement Learning: A Survey

Transformers have significantly impacted domains like natural language processing, computer vision, and robotics, where they improve performance compared to other neural networks. This survey explores how transformers are used in reinforcement learning (RL), where they are seen as a promising solution for addressing challenges such as unstable training, credit assignment, lack of interpretability, and partial observability. We begin by providing a brief domain overview of RL, followed by a discussion on the challenges of classical RL algorithms. Next, we delve into the properties of the transformer and its variants and discuss the characteristics that make them well-suited to address the challenges inherent in RL. We examine the application of transformers to various aspects of RL, including representation learning, transition and reward function modeling, and policy optimization. We also discuss recent research that aims to enhance the interpretability and efficiency of transformers in RL, using visualization techniques and efficient training strategies. Often, the transformer architecture must be tailored to the specific needs of a given application. We present a broad overview of how transformers have been adapted for several applications, including robotics, medicine, language modeling, cloud computing, and combinatorial optimization. We conclude by discussing the limitations of using transformers in RL and assess their potential for catalyzing future breakthroughs in this field.

cs.LG↗

arXiv2024

Deep Learning for Melt Pool Depth Contour Prediction From Surface Thermal Images via Vision Transformers

Insufficient overlap between the melt pools produced during Laser Powder Bed Fusion (L-PBF) can lead to lack-of-fusion defects and deteriorated mechanical and fatigue performance. In-situ monitoring of the melt pool subsurface morphology requires specialized equipment that may not be readily accessible or scalable. Therefore, we introduce a machine learning framework to correlate in-situ two-color thermal images observed via high-speed color imaging to the two-dimensional profile of the melt pool cross-section. Specifically, we employ a hybrid CNN-Transformer architecture to establish a correlation between single bead off-axis thermal image sequences and melt pool cross-section contours measured via optical microscopy. In this architecture, a ResNet model embeds the spatial information contained within the thermal images to a latent vector, while a Transformer model correlates the sequence of embedded vectors to extract temporal information. Our framework is able to model the curvature of the subsurface melt pool structure, with improved performance in high energy density regimes compared to analytical melt pool models. The performance of this model is evaluated through dimensional and geometric comparisons to the corresponding experimental melt pool observations.

cs.LG↗

arXiv2024

Frequency Guidance Matters: Skeletal Action Recognition by Frequency-Aware Mixed Transformer

Recently, transformers have demonstrated great potential for modeling long-term dependencies from skeleton sequences and thereby gained ever-increasing attention in skeleton action recognition. However, the existing transformer-based approaches heavily rely on the naive attention mechanism for capturing the spatiotemporal features, which falls short in learning discriminative representations that exhibit similar motion patterns. To address this challenge, we introduce the Frequency-aware Mixed Transformer (FreqMixFormer), specifically designed for recognizing similar skeletal actions with subtle discriminative motions. First, we introduce a frequency-aware attention module to unweave skeleton frequency representations by embedding joint features into frequency attention maps, aiming to distinguish the discriminative movements based on their frequency coefficients. Subsequently, we develop a mixed transformer architecture to incorporate spatial features with frequency features to model the comprehensive frequency-spatial patterns. Additionally, a temporal transformer is proposed to extract the global correlations across frames. Extensive experiments show that FreqMiXFormer outperforms SOTA on 3 popular skeleton action recognition datasets, including NTU RGB+D, NTU RGB+D 120, and NW-UCLA datasets.

cs.CV↗

arXiv2024

Deep Learning-Driven Microstructure Characterization and Vickers Hardness Prediction of Mg-Gd Alloys

In the field of materials science, exploring the relationship between composition, microstructure, and properties has long been a critical research focus. The mechanical performance of solid-solution Mg-Gd alloys is significantly influenced by Gd content, dendritic structures, and the presence of secondary phases. To better analyze and predict the impact of these factors, this study proposes a multimodal fusion learning framework based on image processing and deep learning techniques. This framework integrates both elemental composition and microstructural features to accurately predict the Vickers hardness of solid-solution Mg-Gd alloys. Initially, deep learning methods were employed to extract microstructural information from a variety of solid-solution Mg-Gd alloy images obtained from literature and experiments. This provided precise grain size and secondary phase microstructural features for performance prediction tasks. Subsequently, these quantitative analysis results were combined with Gd content information to construct a performance prediction dataset. Finally, a regression model based on the Transformer architecture was used to predict the Vickers hardness of Mg-Gd alloys. The experimental results indicate that the Transformer model performs best in terms of prediction accuracy, achieving an R^2 value of 0.9. Additionally, SHAP analysis identified critical values for four key features affecting the Vickers hardness of Mg-Gd alloys, providing valuable guidance for alloy design. These findings not only enhance the understanding of alloy performance but also offer theoretical support for future material design and optimization.

cs.LG↗

arXiv2024

High-Fidelity Document Stain Removal via A Large-Scale Real-World Dataset and A Memory-Augmented Transformer

Document images are often degraded by various stains, significantly impacting their readability and hindering downstream applications such as document digitization and analysis. The absence of a comprehensive stained document dataset has limited the effectiveness of existing document enhancement methods in removing stains while preserving fine-grained details. To address this challenge, we construct StainDoc, the first large-scale, high-resolution ($2145\times2245$) dataset specifically designed for document stain removal. StainDoc comprises over 5,000 pairs of stained and clean document images across multiple scenes. This dataset encompasses a diverse range of stain types, severities, and document backgrounds, facilitating robust training and evaluation of document stain removal algorithms. Furthermore, we propose StainRestorer, a Transformer-based document stain removal approach. StainRestorer employs a memory-augmented Transformer architecture that captures hierarchical stain representations at part, instance, and semantic levels via the DocMemory module. The Stain Removal Transformer (SRTransformer) leverages these feature representations through a dual attention mechanism: an enhanced spatial attention with an expanded receptive field, and a channel attention captures channel-wise feature importance. This combination enables precise stain removal while preserving document content integrity. Extensive experiments demonstrate StainRestorer's superior performance over state-of-the-art methods on the StainDoc dataset and its variants StainDoc\_Mark and StainDoc\_Seal, establishing a new benchmark for document stain removal. Our work highlights the potential of memory-augmented Transformers for this task and contributes a valuable dataset to advance future research.

cs.CV↗

arXiv2025

Utilizing Evolution Strategies to Train Transformers in Reinforcement Learning

We explore the capability of evolution strategies to train an agent with a policy based on a transformer architecture in a reinforcement learning setting. We performed experiments using OpenAI's highly parallelizable evolution strategy to train Decision Transformer in the MuJoCo Humanoid locomotion environment and in the environment of Atari games, testing the ability of this black-box optimization technique to train even such relatively large and complicated models (compared to those previously tested in the literature). The examined evolution strategy proved to be, in general, capable of achieving strong results and managed to produce high-performing agents, showcasing evolution's ability to tackle the training of even such complex models.

cs.LG↗

arXiv2025

Architectural and Inferential Inductive Biases For Exchangeable Sequence Modeling

Autoregressive models have emerged as a powerful framework for modeling exchangeable sequences - i.i.d. observations when conditioned on some latent factor - enabling direct modeling of uncertainty from missing data (rather than a latent). Motivated by the critical role posterior inference plays as a subroutine in decision-making (e.g., active learning, bandits), we study the inferential and architectural inductive biases that are most effective for exchangeable sequence modeling. For the inference stage, we highlight a fundamental limitation of the prevalent single-step generation approach: inability to distinguish between epistemic and aleatoric uncertainty. Instead, a long line of works in Bayesian statistics advocates for multi-step autoregressive generation; we demonstrate this "correct approach" enables superior uncertainty quantification that translates into better performance on downstream decision-making tasks. This naturally leads to the next question: which architectures are best suited for multi-step inference? We identify a subtle yet important gap between recently proposed Transformer architectures for exchangeable sequences (Muller et al., 2022; Nguyen & Grover, 2022; Ye & Namkoong, 2024), and prove that they in fact cannot guarantee exchangeability despite introducing significant computational overhead. We illustrate our findings using controlled synthetic settings, demonstrating how custom architectures can significantly underperform standard causal masks, underscoring the need for new architectural innovations.

cs.LG↗

arXiv2025

Multi Source COVID-19 Detection via Kernel-Density-based Slice Sampling

We present our solution for the Multi-Source COVID-19 Detection Challenge, which classifies chest CT scans from four distinct medical centers. To address multi-source variability, we employ the Spatial-Slice Feature Learning (SSFL) framework with Kernel-Density-based Slice Sampling (KDS). Our preprocessing pipeline combines lung region extraction, quality control, and adaptive slice sampling to select eight representative slices per scan. We compare EfficientNet and Swin Transformer architectures on the validation set. The EfficientNet model achieves an F1-score of 94.68%, compared to the Swin Transformer's 93.34%. The results demonstrate the effectiveness of our KDS-based pipeline on multi-source data and highlight the importance of dataset balance in multi-institutional medical imaging evaluation.

eess.IV↗

arXiv2025

Large Language Models Inference Engines based on Spiking Neural Networks

Foundational models based on the transformer architecture are currently the state-of-the-art in general language modeling, as well as in scientific areas such as material science and climate. However, training and deploying these models is computationally challenging as the time and space complexity has a quadratic relation to the input sequence length. Several efforts exploring efficient computational paradigms and model architectures to address these limitations have been made. In this work, we explore spiking neural networks (SNNs) to design transformer models. A challenge in training large-scale SNNs, using existing surrogate learning methods is inefficient and time-consuming. On the other hand, techniques to convert existing transformer-based models to their SNN equivalent are not scalable, as achieving optimal performance comes at the cost of a large number of spike time-steps, i.e. increased latency. To address this, we propose NeurTransformer, a methodology for designing transformer-based SNN for inference using a supervised fine-tuning approach with existing conversion methods. The proposed methodology works by: (1) replacing the self-attention mechanism with a spike-based self-attention (SSA), (2) converting the feed-forward block of the trained transformer model to its equivalent SNN, and (3) fine-tuning the SSA block using SNN-based surrogate learning algorithms. We benchmark the proposed methodology and demonstrate its accuracy and scalability using three variants of the GPT-2 model of increasing model size. We observe that the converted GPT-2 small models demonstrate a 5-12% loss in cosine similarity and a 9.7% reduction in perplexity. Finally, we demonstrate the energy efficiency of the SSA block compared to the ASA block and show between 64.71% and 85.28% reductions in estimated energy consumption when implementing the self-attention mechanism on a digital hardware.

cs.LG↗

arXiv2025

Hybrid CNN-Transformer Based Sparse Channel Prediction for High-Mobility OTFS Systems

High-mobility scenarios in next-generation wireless networks, such as those involving vehicular communications, require ultra-reliable and low-latency communications (URLLC). However, rapidly time-varying channels pose significant challenges to traditional OFDM-based systems due to the Doppler effect and channel aging. Orthogonal time frequency space (OTFS) modulation offers resilience by representing channels in the quasi-static delay-Doppler (DD) domain. This letter proposes a novel channel prediction framework for OTFS systems using a hybrid convolutional neural network and transformer (CNN-Transformer) architecture. The CNN extracts compact features that exploit the DD-domain sparsity of the channel matrices, while the transformer models temporal dependencies with causal masking for consistency. Simulation experiments under extreme $500$ \si{km/h} mobility conditions demonstrate that the proposed method outperforms state-of-the-art baselines, reducing the root mean square error and mean absolute error by $12.2\%$ and $9.4\%$, respectively. These results demonstrate the effectiveness of DD-domain representations and the proposed model in accurately predicting channels in high-mobility scenarios, thereby supporting the stringent URLLC requirements in future wireless systems.

cs.IT↗

arXiv2025

A cross-modal pre-training framework with video data for improving performance and generalization of distributed acoustic sensing

Fiber-optic distributed acoustic sensing (DAS) has emerged as a critical Internet-of-Things (IoT) sensing technology with broad industrial applications. However, the two-dimensional spatial-temporal morphology of DAS signals presents analytical challenges where conventional methods prove suboptimal, while being well-suited for deep learning approaches. Although our previous work, DAS Masked Autoencoder (DAS-MAE), established state-of-the-art performance and generalization without labels, it is not satisfactory in frequency analysis in temporal-dominated DAS data. Moreover, the limitation of effective training data fails to address the substantial data requirements inherent to Transformer architectures in DAS-MAE. To overcome these limitations, we present an enhanced framework incorporating short-time Fourier transform (STFT) for explicit temporal-frequency feature extraction and pioneering video-to-DAS cross-modal pre-training to mitigate data constraints. This approach learns high-level representations (e.g., event classification) through label-free reconstruction tasks. Experimental results demonstrate transformative improvements: 0.1% error rate in few-shot classification (90.9% relative improvement over DAS-MAE) and 4.7% recognition error in external damage prevention applications (75.4% improvement over from-scratch training). As the first work to pioneer video-to-DAS cross-modal pre-training, available training resources are expanded by bridging computer vision and distributed sensing areas. The enhanced performance and generalization facilitate DAS deployment across diverse industrial scenarios while advancing cross-modal representation learning for industrial IoT sensing.

eess.SP↗

arXiv2025

Distilled HuBERT for Mobile Speech Emotion Recognition: A Cross-Corpus Validation Study

Speech Emotion Recognition (SER) has significant potential for mobile applications, yet deployment remains constrained by the computational demands of state-of-the-art transformer architectures. This paper presents a mobile-efficient SER system based on DistilHuBERT, a distilled and 8-bit quantized transformer that achieves approximately 92% parameter reduction compared to full-scale Wav2Vec 2.0 models while maintaining competitive accuracy. We conduct a rigorous 5-fold Leave-One-Session-Out (LOSO) cross-validation on the IEMOCAP dataset to ensure speaker independence, augmented with cross-corpus training on CREMA-D to enhance generalization. Cross-corpus training with CREMA-D yields a 1.2% improvement in Weighted Accuracy, a 1.4% gain in Macro F1-score, and a 32% reduction in cross-fold variance, with the Neutral class showing the most substantial benefit at 5.4% F1-score improvement. Our approach achieves an Unweighted Accuracy of 61.4% with a quantized model footprint of only 23 MB, representing approximately 91% of the Unweighted Accuracy of a full-scale baseline. Cross-corpus evaluation on RAVDESS reveals that the theatrical nature of acted emotions causes predictions to cluster by arousal level rather than by specific emotion categories - happiness predictions systematically bleed into anger predictions, and sadness predictions bleed into neutral predictions, due to acoustic saturation when actors prioritize clarity over subtlety. Despite this theatricality effect reducing overall RAVDESS accuracy to 46.64%, the model maintains robust arousal detection with 99% recall for anger, 55% recall for neutral, and 27% recall for sadness. These findings demonstrate a Pareto-optimal tradeoff between model size and accuracy, enabling practical affect recognition on resource-constrained mobile devices.

cs.SD↗

arXiv2026

Cascaded Transformer for Robust and Scalable SLA Decomposition via Amortized Optimization

The evolution toward 6G networks increasingly relies on network slicing to provide tailored, End-to-End (E2E) logical networks over shared physical infrastructures. A critical challenge is effectively decomposing E2E Service Level Agreements (SLAs) into domain-specific SLAs, which current solutions handle through computationally intensive, iterative optimization processes that incur substantial latency and complexity. To address this, we introduce Casformer, a cascaded Transformer architecture designed for fast, optimization-free SLA decomposition. Casformer leverages historical domain feedback encoded through domain-specific Transformer encoders in its first layer, and integrates cross-domain dependencies using a Transformer-based aggregator in its second layer. The model is trained under a learning paradigm inspired by Domain-Informed Neural Networks (DINNs), incorporating risk-informed modeling and amortized optimization to learn a stable, forward-only SLA decomposition policy. Extensive evaluations demonstrate that Casformer achieves improved SLA decomposition quality against state-of-the-art optimization-based frameworks, while exhibiting enhanced scalability and robustness under volatile and noisy network conditions. In addition, its forward-only design reduces runtime complexity and simplifies deployment and maintenance. These insights reveal the potential of combining amortized optimization with Transformer-based sequence modeling to advance network automation, providing a scalable and efficient solution suitable for real-time SLA management in advanced 5G-and-beyond network environments.

cs.NI↗

arXiv2026

Neural Dynamics Self-Attention for Spiking Transformers

Integrating Spiking Neural Networks (SNNs) with Transformer architectures offers a promising pathway to balance energy efficiency and performance, particularly for edge vision applications. However, existing Spiking Transformers face two critical challenges: (i) a substantial performance gap compared to their Artificial Neural Networks (ANNs) counterparts and (ii) high memory overhead during inference. Through theoretical analysis, we attribute both limitations to the Spiking Self-Attention (SSA) mechanism: the lack of locality bias and the need to store large attention matrices. Inspired by the localized receptive fields (LRF) and membrane-potential dynamics of biological visual neurons, we propose LRF-Dyn, which uses spiking neurons with localized receptive fields to compute attention while reducing memory requirements. Specifically, we introduce a LRF method into SSA to assign higher weights to neighboring regions, strengthening local modeling and improving performance. Building on this, we approximate the resulting attention computation via charge-fire-reset dynamics, eliminating explicit attention-matrix storage and reducing inference-time memory. Extensive experiments on visual tasks confirm that our method reduces memory overhead while delivering significant performance improvements. These results establish it as a key unit for achieving energy-efficient Spiking Transformers.

cs.NE↗

arXiv2026

TEMPO: Transformers for Temporal Disease Progression from Cross-Sectional Data

Event-Based Models (EBMs) infer biomarker progression from cross-sectional data but typically only as ordinal sequences and rely on rigid model assumptions. We propose \textsc{Tempo}, a Transformer architecture that learns both ordinal and continuous event sequences through simulation-based supervised learning. \textsc{Tempo} uses two Transformer modules: one treats biomarkers as tokens to infer event sequencing; the other treats patients as tokens, representing each by their per-biomarker abnormality profile, to infer patients' disease stages. On synthetic benchmarks, \textsc{Tempo} reduces normalized Kendall's Tau distance by 52.89\% and staging MAE by 25.33\% compared to state-of-the-art SA-EBM, with larger reductions in high-dimensional settings (58.88\% and 61.10\%). Applied to ADNI, \textsc{Tempo} recovers a biologically plausible Alzheimer's progression: early medial temporal atrophy, followed by amyloid accumulation and cognitive decline, and late-stage tau pathology with terminal acceleration of global neurodegeneration -- broadly consistent with established disease models. \textsc{Tempo} also eliminates the need to derive custom inference algorithms and enables rapid empirical comparison of generative hypotheses.

cs.LG↗