The recently proposed xLSTM is a powerful model that leverages expressive multiplicative gating and residual connections, providing the temporal capacity needed for long-horizon forecasting and representation learning. This architecture has demonstrated success in time series forecasting, lossless compression, and even large-scale language modeling tasks, where its linear memory footprint and fast inference make it a viable alternative to Transformers. Despite its growing popularity, no prior work has explored xLSTM for anomaly detection. In this work, we fill this gap by proposing xLSTMAD, the first anomaly detection method that integrates a full encoder-decoder xLSTM architecture, purpose-built for multivariate time series data. Our encoder processes input sequences to capture historical context, while the decoder is devised in two separate variants of the method. In the forecasting approach, the decoder iteratively generates forecasted future values xLSTMAD-F, while the reconstruction approach reconstructs the input time series from its encoded counterpart xLSTMAD-R. We investigate the performance of two loss functions: Mean Squared Error (MSE), and Soft Dynamic Time Warping (SoftDTW) to consider local reconstruction fidelity and global sequence alignment, respectively. We evaluate our method on the comprehensive TSB-AD-M benchmark, which spans 17 real-world datasets, using state-of-the-art challenging metrics such as VUS-PR. In our results, xLSTM showcases state-of-the-art accuracy, outperforming 23 popular anomaly detection baselines. Our paper is the first work revealing the powerful modeling capabilities of xLSTM for anomaly detection, paving the way for exciting new developments on this subject. Our code is available at: https://github.com/Nyderx/xlstmad

Recent advancements in Recurrent Neural Network (RNN) architectures, particularly the Extended Long Short Term Memory (xLSTM), have addressed the limitations of traditional Long Short Term Memory (LSTM) networks by introducing exponential gating and enhanced memory structures. These improvements make xLSTM suitable for time-series prediction tasks as they exhibit the ability to model long-term temporal dependencies better than LSTMs. Despite their potential, these xLSTM-based models remain largely unexplored in the context of vehicle trajectory prediction. Therefore, this paper introduces a novel xLSTM-based vehicle trajectory prediction framework, X-TRAJ, and its physics-aware variant, X-TRACK (eXtended LSTM for TRAjectory prediction Constraint by Kinematics), which explicitly integrates vehicle motion kinematics into the model learning process. By introducing physical constraints, the proposed model generates realistic and feasible trajectories. A comprehensive evaluation on the highD and NGSIM datasets demonstrates that X-TRACK outperforms state-of-the-art baselines.

Toxic comment detection remains a challenging task, where transformer-based models (e.g., BERT) incur high computational costs and degrade on minority toxicity classes, while classical ensembles lack semantic adaptability. We propose xLSTM, a parameter-efficient and theoretically grounded framework that unifies cosine-similarity gating, adaptive feature prioritization, and principled class rebalancing. A learnable reference vector {v} in {R}^d modulates contextual embeddings via cosine similarity, amplifying toxic cues and attenuating benign signals to yield stronger gradients under severe class imbalance. xLSTM integrates multi-source embeddings (GloVe, FastText, BERT CLS) through a projection layer, a character-level BiLSTM for morphological cues, embedding-space SMOTE for minority augmentation, and adaptive focal loss with dynamic class weighting. On the Jigsaw Toxic Comment benchmark, xLSTM attains 96.0% accuracy and 0.88 macro-F1, outperforming BERT by 33% on threat and 28% on identity_hate categories, with 15 times fewer parameters and 50ms inference latency. Cosine gating contributes a +4.8% F1 gain in ablations. The results establish a new efficiency adaptability frontier, demonstrating that lightweight, theoretically informed architectures can surpass large pretrained models on imbalanced, domain-specific NLP tasks.

Since then, LSTMs have stood the test of time and contributed to numerous deep learning success stories, in particular they constituted the first Large Language Models (LLMs). However, the advent of the Transformer technology with parallelizable self-attention at its core marked the dawn of a new era, outpacing LSTMs at scale.

With the growing reliance on the vulnerable Automatic Dependent Surveillance-Broadcast (ADS-B) protocol in air traffic management (ATM), ensuring security is critical. This study investigates emerging machine learning models and training strategies to improve AI-based intrusion detection systems (IDS) for ADS-B. Focusing on ground-based ATM systems, we evaluate two deep learning IDS implementations: one using a transformer encoder and the other an extended Long Short-Term Memory (xLSTM) network, marking the first xLSTM-based IDS for ADS-B. A transfer learning strategy was employed, involving pre-training on benign ADS-B messages and fine-tuning with labeled data containing instances of tampered messages. Results show this approach outperforms existing methods, particularly in identifying subtle attacks that progressively undermine situational awareness. The xLSTM-based IDS achieves an F1-score of 98.9%, surpassing the transformer-based model at 94.3%. Tests on unseen attacks validated the generalization ability of the xLSTM model. Inference latency analysis shows that the 7.26-second delay introduced by the xLSTM-based IDS fits within the Secondary Surveillance Radar (SSR) refresh interval (5-12 s), although it may be restrictive for time-critical operations. While the transformer-based IDS achieves a 2.1-second latency, it does so at the cost of lower detection performance.

In recent years, self-supervised learning has amassed significant interest for training deep neural representations without labeled data. One such self-supervised learning approach is masked spectrogram modeling, where the objective is to learn semantically rich contextual representations by predicting removed or hidden portions of the input audio spectrogram. With the Transformer neural architecture at its core, masked spectrogram modeling has emerged as the prominent approach for learning general purpose audio representations, a.k.a. audio foundation models. Meanwhile, addressing the issues of the Transformer architecture, in particular the underlying Scaled Dot-product Attention operation, which scales quadratically with input sequence length, has led to renewed interest in recurrent sequence modeling approaches. Among them, Selective structured state space models (such as Mamba) and extended Long Short-Term Memory (xLSTM) are the two most promising approaches which have experienced widespread adoption. While the body of work on these two topics continues to grow, there is currently a lack of an adequate overview encompassing the intersection of these topics. In this paper, we present a comprehensive overview of the aforementioned research domains, covering masked spectrogram modeling and the previously mentioned neural sequence modeling architectures, Mamba and xLSTM. Further, we compare Transformers, Mamba and xLSTM based masked spectrogram models in a unified, reproducible framework on ten diverse downstream audio classification tasks, which will help interested readers to make informed decisions regarding suitability of the evaluated approaches to adjacent applications.

This study investigates the application of novel model architectures and large-scale foundational models in temporal series analysis of lower limb rehabilitation motion data, aiming to leverage advancements in machine learning and artificial intelligence to empower active rehabilitation guidance strategies for post-stroke patients in limb motor function recovery. Utilizing the SIAT-LLMD dataset of lower limb movement data proposed by the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, we systematically elucidate the implementation and analytical outcomes of the innovative xLSTM architecture and the foundational model Lag-Llama in short-term temporal prediction tasks involving joint kinematics and dynamics parameters. The research provides novel insights for AI-enabled medical rehabilitation applications, demonstrating the potential of cutting-edge model architectures and large-scale models in rehabilitation medicine temporal prediction. These findings establish theoretical foundations for future applications of personalized rehabilitation regimens, offering significant implications for the development of customized therapeutic interventions in clinical practice.

In the 1990s, the constant error carousel and gating were introduced as the central ideas of the Long Short-Term Memory (LSTM). Since then, LSTMs have stood the test of time and contributed to numerous deep learning success stories, in particular they constituted the first Large Language Models (LLMs). However, the advent of the Transformer technology with parallelizable self-attention at its core marked the dawn of a new era, outpacing LSTMs at scale. We now raise a simple question: How far do we get in language modeling when scaling LSTMs to billions of parameters, leveraging the latest techniques from modern LLMs, but mitigating known limitations of LSTMs? Firstly, we introduce exponential gating with appropriate normalization and stabilization techniques.

However, the quadratic complexityO ( n 2) of the attention mechanism (wheren is the sequence length) makes it computationally expensive to train and deploy large models, particularly for long sequences. This inherent limitation poses significant challenges for scalability and efficiency in real-world applications. One highly effective technique widely adopted to mitigate these challenges in training and deploying such massive models is the Mixture of Experts (MoE) framework [5, 11]. By design, in a MoE architecture, at inference time, the model intelligently utilizes only a sparse subset of its total parameters to process each input, leading to a dramatic reduction in the computational requirements at runtime and enabling more efficient scaling. The sparse MoE approach has been successfully applied to various models, demonstrating significant improvements in efficiency while maintaining or even enhancing performance [2]. Traditional Long Short-Term Memory (LSTM) networks, while demonstrably powerful in sequence modeling, inherently struggle with effectively managing long-term dependencies and achieving efficient associative recall, particularly when dealing with extended sequences. The Extended Long Short-Term Memory (xLSTM) architecture [1] directly addresses these fundamental limitations by introducing novel memory structures and optimized computation approaches within the LSTM unit itself.