RNN
-
Learning State-Tracking from Code Using Linear RNNs
2026-02-16 15:07:51Julien Siems, Riccardo Grazzi, Kirill Kalinin, Hitesh Ballani, Babak RahmaniAbstract
Over the last years, state-tracking tasks, particularly permutation composition, have become a testbed to understand the limits of sequence models architectures like Transformers and RNNs (linear and non-linear). However, these are often sequence-to-sequence tasks: learning to map actions (permutations) to states, which is incompatible with the next-token prediction setting commonly used to train language models. We address this gap by converting permutation composition into code via REPL traces that interleave state-reveals through prints and variable transformations. We show that linear RNNs capable of state-tracking excel also in this setting, while Transformers still fail. Motivated by this representation, we investigate why tracking states in code is generally difficult: actions are not always fully observable. We frame this as tracking the state of a probabilistic finite-state automaton with deterministic state reveals and show that linear RNNs can be worse than non-linear RNNs at tracking states in this setup.
Abstract (translated)
URL
https://arxiv.org/abs/2602.14814
PDF
-
Cross-household Transfer Learning Approach with LSTM-based Demand Forecasting
2026-02-15 18:35:14Manal Rahal, Bestoun S. Ahmed, Roger Renstr\"om, Robert StenerAbstract
With the rapid increase in residential heat pump (HP) installations, optimizing hot water production in households is essential, yet it faces major technical and scalability challenges. Adapting production to actual household needs requires accurate forecasting of hot water demand to ensure comfort and, most importantly, to reduce energy waste. However, the conventional approach of training separate machine learning models for each household becomes computationally expensive at scale, particularly in cloud-connected HP deployments. This study introduces DELTAiF, a transfer learning (TL) based framework that provides scalable and accurate prediction of household hot water consumption. By predicting large hot water usage events, such as showers, DELTAiF enables adaptive yet scalable hot water production at the household level. DELTAiF leverages learned knowledge from a representative household and fine-tunes it across others, eliminating the need to train separate machine learning models for each HP installation. This approach reduces overall training time by approximately 67 percent while maintaining high predictive accuracy values between 0.874 and 0.991, and mean absolute percentage error values between 0.001 and 0.017. The results show that TL is particularly effective when the source household exhibits regular consumption patterns, enabling hot water demand forecasting at scale.
Abstract (translated)
URL
https://arxiv.org/abs/2602.14267
PDF
-
Named Entity Recognition for Payment Data Using NLP
2026-02-15 06:28:57Srikumar NayakAbstract
Named Entity Recognition (NER) has emerged as a critical component in automating financial transaction processing, particularly in extracting structured information from unstructured payment data. This paper presents a comprehensive analysis of state-of-the-art NER algorithms specifically designed for payment data extraction, including Conditional Random Fields (CRF), Bidirectional Long Short-Term Memory with CRF (BiLSTM-CRF), and transformer-based models such as BERT and FinBERT. We conduct extensive experiments on a dataset of 50,000 annotated payment transactions across multiple payment formats including SWIFT MT103, ISO 20022, and domestic payment systems. Our experimental results demonstrate that fine-tuned BERT models achieve an F1-score of 94.2% for entity extraction, outperforming traditional CRF-based approaches by 12.8 percentage points. Furthermore, we introduce PaymentBERT, a novel hybrid architecture combining domain-specific financial embeddings with contextual representations, achieving state-of-the-art performance with 95.7% F1-score while maintaining real-time processing capabilities. We provide detailed analysis of cross-format generalization, ablation studies, and deployment considerations. This research provides practical insights for financial institutions implementing automated sanctions screening, anti-money laundering (AML) compliance, and payment processing systems.
Abstract (translated)
URL
https://arxiv.org/abs/2602.14009
PDF
-
Synaptic Activation and Dual Liquid Dynamics for Interpretable Bio-Inspired Models
2026-02-13 15:23:37M\'onika Farsang, Radu GrosuAbstract
In this paper, we present a unified framework for various bio-inspired models to better understand their structural and functional differences. We show that liquid-capacitance-extended models lead to interpretable behavior even in dense, all-to-all recurrent neural network (RNN) policies. We further demonstrate that incorporating chemical synapses improves interpretability and that combining chemical synapses with synaptic activation yields the most accurate and interpretable RNN models. To assess the accuracy and interpretability of these RNN policies, we consider the challenging lane-keeping control task and evaluate performance across multiple metrics, including turn-weighted validation loss, neural activity during driving, absolute correlation between neural activity and road trajectory, saliency maps of the networks' attention, and the robustness of their saliency maps measured by the structural similarity index.
Abstract (translated)
URL
https://arxiv.org/abs/2602.13017
PDF
-
A two-step approach for speech enhancement in low-SNR scenarios using cyclostationary beamforming and DNNs
2026-02-13 14:59:43Giovanni Bologni, Nicol\'as Arrieta Larraza, Richard Heusdens, Richard C. HendriksAbstract
Deep Neural Networks (DNNs) often struggle to suppress noise at low signal-to-noise ratios (SNRs). This paper addresses speech enhancement in scenarios dominated by harmonic noise and proposes a framework that integrates cyclostationarity-aware preprocessing with lightweight DNN-based denoising. A cyclic minimum power distortionless response (cMPDR) spectral beamformer is used as a preprocessing block. It exploits the spectral correlations of cyclostationary noise to suppress harmonic components prior to learning-based enhancement and does not require modifications to the DNN architecture. The proposed pipeline is evaluated in a single-channel setting using two DNN architectures: a simple and lightweight convolutional recurrent neural network (CRNN), and a state-of-the-art model, namely ultra-low complexity network (ULCNet). Experiments on synthetic data and real-world recordings dominated by rotating machinery noise demonstrate consistent improvements over end-to-end DNN baselines, particularly at low SNRs. Remarkably, a parameter-efficient CRNN with cMPDR preprocessing surpasses the performance of the larger ULCNet operating on raw or Wiener-filtered inputs. These results indicate that explicitly incorporating cyclostationarity as a signal prior is more effective than increasing model capacity alone for suppressing harmonic interference.
Abstract (translated)
URL
https://arxiv.org/abs/2602.12986
PDF
-
PixelRush: Ultra-Fast, Training-Free High-Resolution Image Generation via One-step Diffusion
2026-02-13 09:54:27Hong-Phuc Lai, Phong Nguyen, Anh TranAbstract
Pre-trained diffusion models excel at generating high-quality images but remain inherently limited by their native training resolution. Recent training-free approaches have attempted to overcome this constraint by introducing interventions during the denoising process; however, these methods incur substantial computational overhead, often requiring more than five minutes to produce a single 4K image. In this paper, we present PixelRush, the first tuning-free framework for practical high-resolution text-to-image generation. Our method builds upon the established patch-based inference paradigm but eliminates the need for multiple inversion and regeneration cycles. Instead, PixelRush enables efficient patch-based denoising within a low-step regime. To address artifacts introduced by patch blending in few-step generation, we propose a seamless blending strategy. Furthermore, we mitigate over-smoothing effects through a noise injection mechanism. PixelRush delivers exceptional efficiency, generating 4K images in approximately 20 seconds representing a 10$\times$ to 35$\times$ speedup over state-of-the-art methods while maintaining superior visual fidelity. Extensive experiments validate both the performance gains and the quality of outputs achieved by our approach.
Abstract (translated)
URL
https://arxiv.org/abs/2602.12769
PDF
-
Using Deep Learning to Generate Semantically Correct Hindi Captions
2026-02-13 05:06:30Wasim Akram Khan, Anil Kumar VuppalaAbstract
Automated image captioning using the content from the image is very appealing when done by harnessing the capability of computer vision and natural language processing. Extensive research has been done in the field with a major focus on the English language which gives the scope for further developments in the same with consideration of popular foreign languages. This research utilizes distinct models for translating the image caption into Hindi, the fourth most popular language across the world. Exploring the multi-modal architectures this research comprises local visual features, global visual features, attention mechanisms, and pre-trained models. Using google cloud translator on the image dataset from Flickr8k, Hindi image descriptions have been generated. Pre-trained CNNs like VGG16, ResNet50, and Inception V3 helped in retrieving image characteristics, while the uni-directional and bi-directional techniques of text encoding are used for the text encoding process. An additional Attention layer helps to generate a weight vector and, by multiplying it, combine image characteristics from each time step into a sentence-level feature vector. Bilingual evaluation understudy scores are used to compare the research outcome. Many experiments that serve as a baseline are done for the comparative analysis of the research. An image with a score of BLEU-1 is considered sufficient, whereas one with a score of BLEU-4 is considered to have fluid image captioning. For both BLEU scores, the attention-based bidirectional LSTM with VGG16 produced the best results of 0.59 and 0.19 respectively. The experiments conclude that researchs ability to produce relevant, semantically accurate image captions in Hindi. The research accomplishes the goals and future research can be guided by this research model.
Abstract (translated)
URL
https://arxiv.org/abs/2602.13352
PDF
-
TFT-ACB-XML: Decision-Level Integration of Customized Temporal Fusion Transformer and Attention-BiLSTM with XGBoost Meta-Learner for BTC Price Forecasting
2026-02-12 20:20:56Raiz Ud Din (Artificial Intelligence Lab, Department of Computer Systems Engineering, University of Engineering and Applied Sciences), Saddam Hussain Khan (Interdisciplinary Research Center for Smart Mobility and Logistics, King Fahad University of Petroleum and Minerals)Abstract
Accurate forecasting of Bitcoin (BTC) has always been a challenge because decentralized markets are non-linear, highly volatile, and have temporal irregularities. Existing deep learning models often struggle with interpretability and generalization across diverse market conditions. This research presents a hybrid stacked-generalization framework, TFT-ACB-XML, for BTC closing price prediction. The framework integrates two parallel base learners: a customized Temporal Fusion Transformer (TFT) and an Attention-Customized Bidirectional Long Short-Term Memory network (ACB), followed by an XGBoost regressor as the meta-learner. The customized TFT model handles long-range dependencies and global temporal dynamics via variable selection networks and interpretable single-head attention. The ACB module uses a new attention mechanism alongside the customized BiLSTM to capture short-term sequential dependencies. Predictions from both customized TFT and ACB are weighted through an error-reciprocal weighting strategy. These weights are derived from validation performance, where a model showing lower prediction error receives a higher weight. Finally, the framework concatenates these weighted outputs into a feature vector and feeds the vector to an XGBoost regressor, which captures non-linear residuals and produces the final BTC closing price prediction. Empirical validation using BTC data from October 1, 2014, to January 5, 2026, shows improved performance of the proposed framework compared to recent Deep Learning and Transformer baseline models. The results show a MAPE of 0.65%, an MAE of 198.15, and an RMSE of 258.30 for one-step-ahead out-of-sample under a walk-forward evaluation on the test block. The evaluation period spans the 2024 BTC halving and the spot ETFs (exchange-traded funds) period, which coincide with major liquidity and volatility shifts.
Abstract (translated)
URL
https://arxiv.org/abs/2602.12380
PDF
-
Human-Like Gaze Behavior in Social Robots: A Deep Learning Approach Integrating Human and Non-Human Stimuli
2026-02-12 07:01:17Faezeh Vahedi, Morteza Memari, Ramtin Tabatabaei, Alireza TaheriAbstract
Nonverbal behaviors, particularly gaze direction, play a crucial role in enhancing effective communication in social interactions. As social robots increasingly participate in these interactions, they must adapt their gaze based on human activities and remain receptive to all cues, whether human-generated or not, to ensure seamless and effective communication. This study aims to increase the similarity between robot and human gaze behavior across various social situations, including both human and non-human stimuli (e.g., conversations, pointing, door openings, and object drops). A key innovation in this study, is the investigation of gaze responses to non-human stimuli, a critical yet underexplored area in prior research. These scenarios, were simulated in the Unity software as a 3D animation and a 360-degree real-world video. Data on gaze directions from 41 participants were collected via virtual reality (VR) glasses. Preprocessed data, trained two neural networks-LSTM and Transformer-to build predictive models based on individuals' gaze patterns. In the animated scenario, the LSTM and Transformer models achieved prediction accuracies of 67.6% and 70.4%, respectively; In the real-world scenario, the LSTM and Transformer models achieved accuracies of 72% and 71.6%, respectively. Despite the gaze pattern differences among individuals, our models outperform existing approaches in accuracy while uniquely considering non-human stimuli, offering a significant advantage over previous literature. Furthermore, deployed on the NAO robot, the system was evaluated by 275 participants via a comprehensive questionnaire, with results demonstrating high satisfaction during interactions. This work advances social robotics by enabling robots to dynamically mimic human gaze behavior in complex social contexts.
Abstract (translated)
非言语行为，尤其是目光方向，在社交互动中有效沟通的提升方面扮演着关键角色。随着社交机器人越来越多地参与这些互动，它们必须根据人类活动调整自己的目光，并且对所有线索保持敏感，无论该线索是人为生成还是其它来源的，以确保无缝和有效的交流。本研究旨在增加在各种社交情境下机器人与人眼动行为之间的相似性，包括人类及非人类刺激（例如对话、指向动作、门开关以及物体掉落）。 这项研究的一个关键创新点在于对非人类刺激的眼动反应进行了调查，这是一个以前的研究中相对较少探讨的重要领域。这些场景通过Unity软件作为3D动画和360度真实世界视频被模拟出来。41名参与者的目光方向数据通过虚拟现实（VR）眼镜收集。预处理后的数据训练了两个神经网络模型——LSTM和Transformer——基于个体的眼动模式构建预测模型。 在动画情景中，LSTM和Transformer模型分别实现了67.6%和70.4%的预测准确率；而在真实世界的情景中，它们分别达到了72%和71.6%。尽管个人之间存在眼动模式差异，我们的模型在精度上超越了现有方法，并且独特地考虑到了非人类刺激的影响，在此方面具有显著优势。 此外，该系统被部署到NAO机器人上，并通过一份详尽的问卷由275名参与者进行评估，结果显示在互动过程中满意度很高。这项工作推动了社交机器人的发展，使机器人能够动态模仿在复杂社会环境中的人类眼动行为。
URL
https://arxiv.org/abs/2602.11648
PDF
-
Empirical Study of Gaze Behavior in Children and Young Adults Using Deep Neural Networks and Robot Implementation: A Comparative Analysis of Social Situations
2026-02-12 07:01:04Ramtin Tabatabaei, Milad Hosseini, Ali Mohajerzarrinkelk, Ali F. Meghdari, Alireza TaheriAbstract
In a preliminary exploratory study, our goal was to train deep neural network models to mimic children's and/or adults' gaze behavior in certain social situations to reach this objective. Additionally, we aim to identify potential differences in gaze behavior between these two age groups based on our participants' gaze data. Furthermore, we aimed to assess the practical effectiveness of our adult and children models by deploying them on a Nao robot in real-life settings. To achieve this, we first created two video clips, one animation and one live-action, to depict some social situations. Using an eye-tracking device, we collected eye-tracking data from 24 participants, including 12 children and 12 adults. Then, we utilized deep neural networks, specifically LSTM and Transformer Networks, to analyze and model the gaze patterns of each group of participants. Our results indicate that when the models attempted to predict people's locations (in the next frame), they had an accuracy in the range of 62%-70% with one attempt, which increased by ~20% when attempted twice (i.e. the two highest-ranked predicted labels as outputs). As expected, the result underscores that gaze behavior is not a wholly unique phenomenon. We obtained feedback from 57 new participants to evaluate the robot's functionality. These participants were asked to watch two videos of the robot's performance in each mode and then complete a comprehensive questionnaire. The questionnaire results indicate that the participants expressed satisfaction with the robot's interaction, including its attention, intelligence, and responsiveness to human actions. However, they did not perceive the robot as a social companion comparable to a human. This exploratory study tries to address/show potentials of the social acceptance of robots based on human nonverbal behavioral cues for future research.
Abstract (translated)
在初步探索性研究中，我们的目标是训练深度神经网络模型来模仿儿童和/或成人特定社交情景中的注视行为。此外，我们还旨在基于参与者的眼动数据识别这两个年龄段之间潜在的注视行为差异。另外，为了评估我们的成人和儿童模型的实际效果，我们将它们部署在一个Nao机器人上进行实地测试。为此，我们首先创建了两个视频片段，一个是动画，另一个是实拍镜头，以描绘一些社交情景。使用眼动追踪设备，我们从24名参与者（包括12名儿童和12名成人）收集了眼动数据。然后，我们利用深度神经网络（特别是LSTM和Transformer Networks），来分析并建模每个组参与者的注视模式。 我们的结果显示，在模型尝试预测人们的视线位置（下一个帧中的目标对象）时，一次尝试的准确率在62%-70%之间，而两次尝试（即输出两个最高排名的预测标签）后准确率提高了约20%。如预期的那样，这些结果表明注视行为并不是一种完全独特的现象。 我们从57名新参与者那里获得了反馈以评估机器人的功能。这些参与者的任务是观看机器人在每种模式下表演的两段视频，并完成一份详细的问卷调查。问卷结果显示，参与者对机器人的互动表示满意，包括其注意力、智能性和对人类行为的响应性。然而，他们并没有将机器人视为与人相当的社会伙伴。 这项探索性研究试图基于人类非语言行为线索来展示未来研究中关于机器人社会接受潜力的可能性。
URL
https://arxiv.org/abs/2603.00074
PDF
-
An Efficient Hybrid Deep Learning Approach for Detecting Online Abusive Language
2026-02-12 06:13:16Vuong M. Ngo, Cach N. Dang, Kien V. Nguyen, Mark RoantreeAbstract
The digital age has expanded social media and online forums, allowing free expression for nearly 45% of the global population. Yet, it has also fueled online harassment, bullying, and harmful behaviors like hate speech and toxic comments across social networks, messaging apps, and gaming communities. Studies show 65% of parents notice hostile online behavior, and one-third of adolescents in mobile games experience bullying. A substantial volume of abusive content is generated and shared daily, not only on the surface web but also within dark web forums. Creators of abusive comments often employ specific words or coded phrases to evade detection and conceal their intentions. To address these challenges, we propose a hybrid deep learning model that integrates BERT, CNN, and LSTM architectures with a ReLU activation function to detect abusive language across multiple online platforms, including YouTube comments, online forum discussions, and dark web posts. The model demonstrates strong performance on a diverse and imbalanced dataset containing 77,620 abusive and 272,214 non-abusive text samples (ratio 1:3.5), achieving approximately 99% across evaluation metrics such as Precision, Recall, Accuracy, F1-score, and AUC. This approach effectively captures semantic, contextual, and sequential patterns in text, enabling robust detection of abusive content even in highly skewed datasets, as encountered in real-world scenarios.
Abstract (translated)
数字时代扩展了社交媒体和在线论坛的范围，让全球近45%的人口能够自由表达观点。然而，它也助长了网络骚扰、欺凌以及诸如仇恨言论和有毒评论等有害行为，在社交网络、消息应用程序及游戏社区中广泛存在。研究表明，65%的家长注意到孩子的网上敌对行为，而三分之一的游戏青少年遭受过霸凌。每天有大量的恶意内容在表面网络和暗网论坛中生成并传播。恶意评论的创建者经常使用特定词汇或编码短语来逃避检测，并隐藏其意图。 为应对这些挑战，我们提出了一种结合BERT、CNN（卷积神经网络）和LSTM（长短期记忆网络）架构，并采用ReLU激活函数的混合深度学习模型，以在YouTube评论、在线论坛讨论以及暗网帖子等多个在线平台上检测恶意语言。该模型在一个包含77,620条恶意内容和272,214条非恶意文本样本（比例为1:3.5）的多样化且不平衡的数据集上表现出色，其精确度、召回率、准确率、F1分数及AUC等评价指标均达到了约99%。此方法能够有效捕捉到文本中的语义、上下文和序列模式，即使在现实世界中常见的高度偏斜数据集中也能稳健地检测出恶意内容。
URL
https://arxiv.org/abs/2603.09984
PDF
-
Developing Neural Network-Based Gaze Control Systems for Social Robots
2026-02-11 15:30:39Ramtin Tabatabaei, Alireza TaheriAbstract
During multi-party interactions, gaze direction is a key indicator of interest and intent, making it essential for social robots to direct their attention appropriately. Understanding the social context is crucial for robots to engage effectively, predict human intentions, and navigate interactions smoothly. This study aims to develop an empirical motion-time pattern for human gaze behavior in various social situations (e.g., entering, leaving, waving, talking, and pointing) using deep neural networks based on participants' data. We created two video clips-one for a computer screen and another for a virtual reality headset-depicting different social scenarios. Data were collected from 30 participants: 15 using an eye-tracker and 15 using an Oculus Quest 1 headset. Deep learning models, specifically Long Short-Term Memory (LSTM) and Transformers, were used to analyze and predict gaze patterns. Our models achieved 60% accuracy in predicting gaze direction in a 2D animation and 65% accuracy in a 3D animation. Then, the best model was implemented onto the Nao robot; and 36 new participants evaluated its performance. The feedback indicated overall satisfaction, with those experienced in robotics rating the models more favorably.
Abstract (translated)
在多主体互动中，目光的方向是兴趣和意图的关键指标，使社会机器人能够适当关注这一点变得至关重要。理解社交环境对于机器人的有效参与、预测人类意图以及顺畅地进行交互来说是至关重要的。这项研究旨在使用基于参与者数据的深度神经网络开发出一套经验性运动时间模式，以描述人在各种社交情况下的目光行为（例如：进入、离开、挥手、对话和指物）。我们创建了两个视频片段——一个用于计算机屏幕，另一个用于虚拟现实头盔——来展示不同的社交场景。从30名参与者那里收集数据：15人使用眼动仪，另外15人则使用Oculus Quest 1 头显。研究中运用了深度学习模型（特别是长短期记忆网络LSTM和Transformer）来分析并预测目光模式。我们的模型在2D动画中的目光方向预测准确率为60%，而在3D动画中的准确率达到了65%。然后，最佳模型被部署到了Nao机器人上，并由36名新参与者对其表现进行了评估。反馈表明总体满意度较高，有机器人技术经验的人对该模型的评价更为积极。
URL
https://arxiv.org/abs/2602.10946
PDF
-
I can tell whether you are a Native Hawl^eri Speaker! How ANN, CNN, and RNN perform in NLI-Native Language Identification
Abstract
Native Language Identification (NLI) is a task in Natural Language Processing (NLP) that typically determines the native language of an author through their writing or a speaker through their speaking. It has various applications in different areas, such as forensic linguistics and general linguistics studies. Although considerable research has been conducted on NLI regarding two different languages, such as English and German, the literature indicates a significant gap regarding NLI for dialects and subdialects. The gap becomes wider in less-resourced languages such as Kurdish. This research focuses on NLI within the context of a subdialect of Sorani (Central) Kurdish. It aims to investigate the NLI for Hewlêri, a subdialect spoken in Hewlêr (Erbil), the Capital of the Kurdistan Region of Iraq. We collected about 24 hours of speech by recording interviews with 40 native or non-native Hewlêri speakers, 17 female and 23 male. We created three Neural Network-based models: Artificial Neural Network (ANN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN), which were evaluated through 66 experiments, covering various time-frames from 1 to 60 seconds, undersampling, oversampling, and cross-validation. The RNN model showed the highest accuracy of 95.92% for 5-second audio segmentation, using an 80:10:10 data splitting scheme. The created dataset is the first speech dataset for NLI on the Hewlêri subdialect in the Sorani Kurdish dialect, which can be of benefit to various research areas.
Abstract (translated)
母语识别（NLI）是自然语言处理（NLP）中的一个任务，通常通过作者的写作或说话人的言语来确定其母语。它在法医语言学和一般语言研究等领域有各种应用。尽管有关两种不同语言（如英语和德语）之间的母语识别的研究已经相当广泛，但关于方言及次方言的文献显示了一个显著的研究空白。特别是对于资源较少的语言，例如库尔德语，这一差距更为明显。本项研究专注于在苏莱曼尼亚（Central）库尔德语的一个次方言——胡维里亚（Hewlêri）语境下的母语识别问题。该研究旨在调查居住于伊拉克库尔德斯坦自治区首府胡维里亚的当地居民所使用的胡维里亚次方言的母语识别情况。我们通过录制40名本地或非本地说话者的访谈，收集了大约24小时的语音数据，其中包括17名女性和23名男性。我们创建了三种基于神经网络的模型：人工神经网络（ANN）、卷积神经网络（CNN）以及循环神经网络（RNN），通过66次实验对其进行了评估，这些实验涵盖了从1秒到60秒的不同时间范围、欠采样、过采样及交叉验证。在使用80:10:10的数据分割方案的情况下，RNN模型显示出了最高的95.92%的准确率，该结果来自对5秒钟音频片段的分析。所创建的数据集是首个用于胡维里亚次方言母语识别的语音数据集，在苏莱曼尼亚库尔德语中具有开创性意义，并且可以为各种研究领域带来益处。
URL
https://arxiv.org/abs/2602.10832
PDF
-
When to Memorize and When to Stop: Gated Recurrent Memory for Long-Context Reasoning
2026-02-11 06:14:53Leheng Sheng, Yongtao Zhang, Wenchang Ma, Yaorui Shi, Ting Huang, Xiang Wang, An Zhang, Ke Shen, Tat-Seng ChuaAbstract
While reasoning over long context is crucial for various real-world applications, it remains challenging for large language models (LLMs) as they suffer from performance degradation as the context length grows. Recent work MemAgent has tried to tackle this by processing context chunk-by-chunk in an RNN-like loop and updating a textual memory for final answering. However, this naive recurrent memory update faces two crucial drawbacks: (i) memory can quickly explode because it can update indiscriminately, even on evidence-free chunks; and (ii) the loop lacks an exit mechanism, leading to unnecessary computation after even sufficient evidence is collected. To address these issues, we propose GRU-Mem, which incorporates two text-controlled gates for more stable and efficient long-context reasoning. Specifically, in GRU-Mem, the memory only updates when the update gate is open and the recurrent loop will exit immediately once the exit gate is open. To endow the model with such capabilities, we introduce two reward signals $r^{\text{update}}$ and $r^{\text{exit}}$ within end-to-end RL, rewarding the correct updating and exiting behaviors respectively. Experiments on various long-context reasoning tasks demonstrate the effectiveness and efficiency of GRU-Mem, which generally outperforms the vanilla MemAgent with up to 400\% times inference speed acceleration.
Abstract (translated)
虽然在处理长文本上下文时进行推理对于许多现实世界的应用至关重要，但对于大型语言模型（LLMs）来说却是一个挑战。随着上下文长度的增加，这些模型的表现会逐渐下降。最近的工作MemAgent试图通过采用类似RNN的方式逐段处理上下文，并更新文本内存来最终提供答案，以此解决这一问题。然而，这种简单的递归记忆更新方式存在两个关键缺陷：（i）由于它可能在缺乏依据的情况下更新信息，因此记忆可能会迅速膨胀；（ii）循环中缺少退出机制，在收集到足够证据后仍进行不必要的计算。 为了解决这些问题，我们提出了GRU-Mem模型，该模型通过引入两种由文本控制的门来实现更稳定和高效的长上下文推理。具体来说，在GRU-Mem中，内存仅在更新门开启时才会被更新，并且一旦退出门开启，递归循环将立即结束。 为了赋予模型这些能力，我们提出了两个奖励信号$r^{\text{update}}$和$r^{\text{exit}}$，在端到端的强化学习框架内给予正确的更新行为和退出行为相应的奖励。实验表明，在各种长上下文推理任务上，GRU-Mem的有效性和效率均优于原始的MemAgent模型，并且推理速度提高了高达400%。
URL
https://arxiv.org/abs/2602.10560
PDF
-
Instruct2Act: From Human Instruction to Actions Sequencing and Execution via Robot Action Network for Robotic Manipulation
2026-02-10 16:25:39Archit Sharma, Dharmendra Sharma, John Rebeiro, Peeyush Thakur, Narendra Dhar, Laxmidhar BeheraAbstract
Robots often struggle to follow free-form human instructions in real-world settings due to computational and sensing limitations. We address this gap with a lightweight, fully on-device pipeline that converts natural-language commands into reliable manipulation. Our approach has two stages: (i) the instruction to actions module (Instruct2Act), a compact BiLSTM with a multi-head-attention autoencoder that parses an instruction into an ordered sequence of atomic actions (e.g., reach, grasp, move, place); and (ii) the robot action network (RAN), which uses the dynamic adaptive trajectory radial network (DATRN) together with a vision-based environment analyzer (YOLOv8) to generate precise control trajectories for each sub-action. The entire system runs on a modest system with no cloud services. On our custom proprietary dataset, Instruct2Act attains 91.5% sub-actions prediction accuracy while retaining a small footprint. Real-robot evaluations across four tasks (pick-place, pick-pour, wipe, and pick-give) yield an overall 90% success; sub-action inference completes in < 3.8s, with end-to-end executions in 30-60s depending on task complexity. These results demonstrate that fine-grained instruction-to-action parsing, coupled with DATRN-based trajectory generation and vision-guided grounding, provides a practical path to deterministic, real-time manipulation in resource-constrained, single-camera settings.
Abstract (translated)
机器人在现实环境中往往难以根据自由形式的人类指令执行任务，这主要是由于计算能力和感知能力的限制。我们提出了一种轻量级、完全在设备上运行的管道系统，可以将自然语言命令转换为可靠的操纵动作。我们的方法包括两个阶段： (i) 指令到动作模块（Instruct2Act）：这是一个紧凑型BiLSTM模型结合多头注意力自动编码器，能够解析指令，并将其转化为一系列原子动作（例如，到达、抓取、移动和放置）的有序序列。 (ii) 机器人行动网络（RAN）：该阶段使用动态自适应轨迹径向网络（DATRN）与基于视觉环境分析器（YOLOv8）共同工作，为每个子动作生成精确的控制轨迹。 整个系统在没有任何云端服务支持的情况下就可以在一个相对简单的小型设备上运行。在我们定制的专有数据集上，Instruct2Act在保持较小存储需求的同时，实现了91.5%的子动作预测准确率。实机评估显示，在四个任务（取放、倒水、擦拭和递送）中，成功率总体达到了90%，每个子行动推断时间少于3.8秒，而整个操作执行的时间则根据任务复杂度在30到60秒之间。 这些结果表明，细粒度的指令解析与DATRN基础轨迹生成以及视觉引导定位相结合的方法，在资源受限、单摄像头设置中提供了确定性实时操控实现的实际路径。
URL
https://arxiv.org/abs/2602.09940
PDF
-
Wearable environmental sensing to forecast how legged systems will interact with upcoming terrain
2026-02-09 21:22:39Michael D. Murray, James Tung, Richard W. NuckolsAbstract
Computer-vision (CV) has been used for environmental classification during gait and is often used to inform control in assistive systems; however, the ability to predict how the foot will contact a changing environment is underexplored. We evaluated the feasibility of forecasting the anterior-posterior (AP) foot center-of-pressure (COP) and time-of-impact (TOI) prior to foot-strike on a level-ground to stair-ascent transition. Eight subjects wore an RGB-D camera on their right shank and instrumented insoles while performing the task of stepping onto the stairs. We trained a CNN-RNN to forecast the COP and TOI continuously within a 250ms window prior to foot-strike, termed the forecast horizon (FH). The COP mean-absolute-error (MAE) at 150, 100, and 50ms FH was 29.42mm, 26.82, and 23.72mm respectively. The TOI MAE was 21.14, 20.08, and 17.73ms for 150, 100, and 50ms respectively. While torso velocity had no effect on the error in either task, faster toe-swing speeds prior to foot-strike were found to improve the prediction accuracy in the COP case, however, was insignificant in the TOI case. Further, more anterior foot-strikes were found to reduce COP prediction accuracy but did not affect the TOI prediction accuracy. We also found that our lightweight model was capable at running at 60 FPS on either a consumer grade laptop or an edge computing device. This study demonstrates that forecasting COP and TOI from visual data was feasible using a lightweight model, which may have important implications for anticipatory control in assistive systems.
Abstract (translated)
计算机视觉（CV）已被用于步态过程中的环境分类，并常常被用来为辅助系统提供控制信息；然而，预测脚如何与变化的环境接触的能力则鲜少被探索。我们在本研究中评估了在平地过渡到上楼梯时，在足部着地之前预测前后方向（AP）的足底压力中心（COP）和触地时刻（TOI）的可能性。八个受试者在其右小腿佩戴RGB-D相机，并在执行上楼任务的同时穿有仪器化的鞋垫内衬。我们训练了一个卷积神经网络-递归神经网络（CNN-RNN），以在步态足部着地前的250毫秒窗口内连续预测COP和TOI，我们将这一时间范围称为预测视界（FH）。150、100和50毫秒FH下的COP平均绝对误差分别为29.42毫米、26.82毫米和23.72毫米。TOI的平均绝对误差在150、100和50毫秒时分别记录为21.14毫秒、20.08毫秒和17.73毫秒。虽然躯干速度对任一任务中的错误没有影响，但在足部着地前较高的趾摆动速度被发现可以提高COP预测的准确性，但对TOI案例的影响不显著。此外，我们还发现了更靠前的足着地点会降低COP的预测准确度，但这不会影响TOI的预测准确度。我们还发现我们的轻量级模型能够在消费级笔记本电脑或边缘计算设备上以每秒60帧的速度运行。这项研究表明，使用轻量级模型从视觉数据中预测COP和TOI是可行的，这可能对辅助系统的预见性控制有重要影响。
URL
https://arxiv.org/abs/2602.09209
PDF
-
$texttt{lrnnx}$: A library for Linear RNNs
2026-02-09 15:48:48Karan Bania, Soham Kalburgi, Manit Tanwar, Dhruthi, Aditya Nagarsekar, Harshvardhan Mestha, Naman Chibber, Raj Deshmukh, Anish Sathyanarayanan, Aarush Rathore, Pratham ChhedaAbstract
Linear recurrent neural networks (LRNNs) provide a structured approach to sequence modeling that bridges classical linear dynamical systems and modern deep learning, offering both expressive power and theoretical guarantees on stability and trainability. In recent years, multiple LRNN-based architectures have been proposed, each introducing distinct parameterizations, discretization schemes, and implementation constraints. However, existing implementations are fragmented across different software frameworks, often rely on framework-specific optimizations, and in some cases require custom CUDA kernels or lack publicly available code altogether. As a result, using, comparing, or extending LRNNs requires substantial implementation effort. To address this, we introduce $\texttt{lrnnx}$, a unified software library that implements several modern LRNN architectures under a common interface. The library exposes multiple levels of control, allowing users to work directly with core components or higher-level model abstractions. $\texttt{lrnnx}$ aims to improve accessibility, reproducibility, and extensibility of LRNN research and applications. We make our code available under a permissive MIT license.
Abstract (translated)
线性递归神经网络（LRNN）提供了一种结构化的方法来建模序列，它连接了经典的线性动态系统和现代深度学习技术，既具有表达能力又能保证稳定性和可训练性的理论保障。近年来，提出了多种基于LRNN的架构，每个架构都引入了不同的参数设置、离散化方案以及实现约束条件。然而，现有的实现方式分散在不同的软件框架中，常常依赖于特定框架的优化，并且有些需要定制CUDA内核或根本没有公开可用的代码。因此，使用、比较或扩展LRNN都需要大量的实现工作。 为解决这一问题，我们引入了$\texttt{lrnnx}$，这是一个统一的软件库，实现了几种现代的LRNN架构，在一个通用接口下进行操作。该库暴露了多个级别的控制，使用户可以直接与核心组件或更高层次的模型抽象进行交互。$\texttt{lrnnx}$旨在提高LRNN研究和应用的可访问性、重现性和扩展性。我们以宽松的MIT许可证提供代码。 --- 这段文本主要介绍了线性递归神经网络（LRNN）及其在现代深度学习中的重要地位，同时指出了当前实现这些模型时面临的一些挑战，并提出了一个名为$\texttt{lrnnx}$的新软件库来解决这些问题，通过统一接口和多层级控制提高了研究的可访问性和重现性。
URL
https://arxiv.org/abs/2602.08810
PDF
-
Large Language Models and Impossible Language Acquisition: 'False Promise' or an Overturn of our Current Perspective towards AI
2026-02-09 09:50:12Ziyan wang, Longlong MaAbstract
In Chomsky's provocative critique "The False Promise of CHATGPT," Large Language Models (LLMs) are characterized as mere pattern predictors that do not acquire languages via intrinsic causal and self-correction structures like humans, therefore are not able to distinguish impossible languages. It stands as a representative in a fundamental challenge to the intellectual foundations of AI, for it integrally synthesizes major issues in methodologies within LLMs and possesses an iconic a priori rationalist perspective. We examine this famous critic from both the perspective in pre-existing literature of linguistics and psychology as well as a research based on an experiment inquiring the capacity of learning both possible and impossible languages among LLMs. We constructed a set of syntactically impossible languages by applying certain transformations to English. These include reversing whole sentences, and adding negation based on word-count parity. Two rounds of controlled experiments were each conducted on GPT-2 small models and long short-term memory (LSTM) models. Statistical analysis (Welch's t-test) shows GPT2 small models underperform in learning all of the impossible languages compared to their performance on the possible language (p<.001). On the other hand, LSTM models' performance tallies with Chomsky's argument, suggesting the irreplaceable role of the evolution of transformer architecture. Based on theoretical analysis and empirical findings, we propose a new vision within Chomsky's theory towards LLMs, and a shift of theoretical paradigm outside Chomsky, from his "rationalist-romantics" paradigm to functionalism and empiricism in LLMs research.
Abstract (translated)
在乔姆斯基（Noam Chomsky）的尖锐批评文章《CHATGPT 的虚假承诺》中，大型语言模型（LLMs）被描述为仅仅是模式预测器，它们不像人类那样通过内在因果和自我纠正机制来学习语言。因此，这些模型无法区分不可能存在的语言。这篇评论代表了对人工智能基础理论的一个根本性挑战，因为它综合了LLM方法论中的主要问题，并具有典型的先验理性主义视角。 我们从现有的语言学和心理学文献以及基于实验的研究两方面审视这一著名的批评。该实验研究调查了大型语言模型（LLMs）学习可能与不可能的语言的能力。我们通过应用特定的变换到英语上，构建了一组语法上不可能的语言。这些变换包括反向整个句子，并根据单词数量的奇偶性添加否定。我们在GPT-2小型模型和长短期记忆（LSTM）模型上分别进行了两轮受控实验。统计分析（Welch t检验）显示，在学习所有不可能语言方面，GPT-2小型模型的表现明显不如在可能的语言上的表现（p<.001）。另一方面，LSTM模型的表现支持了乔姆斯基的论点，表明变换架构演进的独特作用。 基于理论分析和实证发现，我们提出了乔姆斯基理论中关于LLMs的新视角，并建议从他的“理性主义浪漫”范式转向功能主义和经验主义在LLMs研究中的新理论框架。
URL
https://arxiv.org/abs/2602.08437
PDF
-
On Sequence-to-Sequence Models for Automated Log Parsing
Abstract
Log parsing is a critical standard operating procedure in software systems, enabling monitoring, anomaly detection, and failure diagnosis. However, automated log parsing remains challenging due to heterogeneous log formats, distribution shifts between training and deployment data, and the brittleness of rule-based approaches. This study aims to systematically evaluate how sequence modelling architecture, representation choice, sequence length, and training data availability influence automated log parsing performance and computational cost. We conduct a controlled empirical study comparing four sequence modelling architectures: Transformer, Mamba state-space, monodirectional LSTM, and bidirectional LSTM models. In total, 396 models are trained across multiple dataset configurations and evaluated using relative Levenshtein edit distance with statistical significance testing. Transformer achieves the lowest mean relative edit distance (0.111), followed by Mamba (0.145), mono-LSTM (0.186), and bi-LSTM (0.265), where lower values are better. Mamba provides competitive accuracy with substantially lower computational cost. Character-level tokenization generally improves performance, sequence length has negligible practical impact on Transformer accuracy, and both Mamba and Transformer demonstrate stronger sample efficiency than recurrent models. Overall, Transformers reduce parsing error by 23.4%, while Mamba is a strong alternative under data or compute constraints. These results also clarify the roles of representation choice, sequence length, and sample efficiency, providing practical guidance for researchers and practitioners.
Abstract (translated)
日志解析是软件系统中的一个关键标准操作程序，它能够实现监控、异常检测和故障诊断。然而，由于异构的日志格式、训练数据与部署数据之间的分布变化以及基于规则方法的脆弱性，自动化日志解析仍然具有挑战性。本研究旨在系统地评估序列建模架构的选择、表示方式选择、序列长度及可用训练数据量如何影响自动化日志解析的表现和计算成本。 我们进行了一项受控实证研究，比较了四种序列建模架构：Transformer、Mamba状态空间模型、单向LSTM（Long Short-Term Memory）以及双向LSTM模型。总计396个模型在不同的数据集配置下进行了训练，并使用相对Levenshtein编辑距离进行评估，且包含统计显著性检验。 结果表明： - Transformer实现了最低的平均相对编辑距离（0.111），其次是Mamba（0.145）、单向LSTM（0.186）和双向LSTM（0.265）。较低值代表更好的表现。 - Mamba在计算成本大幅降低的情况下提供了竞争力的准确性。 - 字符级别的标记化通常改善了性能，而Transformer的精度受序列长度的影响较小。 - 与循环模型相比，Mamba和Transformer展示了更强的数据样本效率。 总体而言，使用Transformer能够将解析错误减少23.4%，而在数据或计算资源受限的情况下，Mamba是一个强有力的选择。这些结果还阐明了表示选择、序列长度及样本效率的作用，并为研究者和实践者提供了实用指导。
URL
https://arxiv.org/abs/2602.07698
PDF
-
Urban Vibrancy Embedding and Application on Traffic Prediction
2026-02-07 04:34:01Sumin Han, Jisun An, Dongman LeeAbstract
Urban vibrancy reflects the dynamic human activity within urban spaces and is often measured using mobile data that captures floating population trends. This study proposes a novel approach to derive Urban Vibrancy embeddings from real-time floating population data to enhance traffic prediction models. Specifically, we utilize variational autoencoders (VAE) to compress this data into actionable embeddings, which are then integrated with long short-term memory (LSTM) networks to predict future embeddings. These are subsequently applied in a sequence-to-sequence framework for traffic forecasting. Our contributions are threefold: (1) We use principal component analysis (PCA) to interpret the embeddings, revealing temporal patterns such as weekday versus weekend distinctions and seasonal patterns; (2) We propose a method that combines VAE and LSTM, enabling forecasting dynamic urban knowledge embedding; and (3) Our approach improves accuracy and responsiveness in traffic prediction models, including RNN, DCRNN, GTS, and GMAN. This study demonstrates the potential of Urban Vibrancy embeddings to advance traffic prediction and offer a more nuanced analysis of urban mobility.
Abstract (translated)
城市活力反映了城市空间内动态的人类活动，通常使用移动数据来捕捉流动人口趋势来进行衡量。本研究提出了一种新的方法，从实时浮动人口数据中推导出城市活力嵌入（Urban Vibrancy embeddings），以增强交通预测模型的性能。具体来说，我们利用变分自编码器(VAE)将这些数据压缩为可操作的嵌入，并将其与长短期记忆网络(LSTM)结合使用来预测未来的嵌入。然后在序列到序列框架中应用这些未来嵌入进行交通预测。 本研究的主要贡献包括： 1. 我们利用主成分分析(PCA)来解释这些嵌入，揭示了诸如工作日和周末区别以及季节性模式等时间模式。 2. 提出了一种结合VAE和LSTM的方法，能够预报动态的城市知识嵌入。 3. 我们的这种方法提高了包括RNN、DCRNN、GTS及GMAN在内的交通预测模型的准确性和响应能力。 这项研究展示了城市活力嵌入在推进交通预测方面以及对城市流动性进行更细致分析方面的潜力。
URL
https://arxiv.org/abs/2602.21232
PDF
