Lightnews — Scholar-powered news

hochreitersepp.bsky.social @hochreitersepp.bsky.social · 8d

xLSTM for long-term context using short sliding windows: arxiv.org/abs/2509.24552
"SWAX, a hybrid consisting of sliding-window attention and xLSTM."
"SWAX trained with stochastic window sizes significantly outperforms regular window attention both on short and long-context problems."

hochreitersepp.bsky.social @hochreitersepp.bsky.social · 22d

xLSTM shines as an Electrocardiogram (ECG) foundation model: arxiv.org/abs/2509.10151

"xECG achieves superior performance over earlier approaches, defining a new baseline for future ECG foundation models."

xLSTM is perfectly suited for time series prediction as shown by TiRex.

1 2

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Sep 3

xLSTM excels in time series forecasting: arxiv.org/abs/2509.01187 .

Introduces "stochastic xLSTM" (StoxLSTM).

"StoxLSTM consistently outperforms state-of-the-art baselines with better robustness and stronger generalization ability."

We know that xLSTM is king at time series from our TiRex.

2

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jul 29

xLSTM for Cellular Traffic Forecasting: arxiv.org/abs/2507.19513

"Empirical results showed a 23% MAE reduction over the original STN and a 30% improvement on unseen data, highlighting strong generalization."

xLSTM shines again in time series forecasting.

2

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jul 8

xLSTM for Monaural Speech Enhancement: arxiv.org/abs/2507.04368

xLSTM has superior performance vs. Mamba and Transformers but is slower than Mamba.

New Triton kernels: xLSTM is faster than MAMBA at training and inference: arxiv.org/abs/2503.13427 and arxiv.org/abs/2503.14376

1

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jul 5

xLSTM for Aspect-based Sentiment Analysis: arxiv.org/abs/2507.01213

Another success story of xLSTM. MEGA: xLSTM with Multihead Exponential Gated Fusion.

Experiments on 3 benchmarks show that MEGA outperforms state-of-the-art baselines with superior accuracy and efficiency”

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jul 1

xLSTM for multivariate time series anomaly detection: arxiv.org/abs/2506.22837

“In our results, xLSTM showcases state-of-the-art accuracy, outperforming 23 popular anomaly detection baselines.”

Again, xLSTM excels in time series analysis.

2 4

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jun 12

xLSTM for Human Action Segmentation: arxiv.org/abs/2506.09650

"HopaDIFF, leveraging a novel cross-input gate attentional xLSTM to enhance holistic-partial long-range reasoning"

"HopaDIFF achieves state-of-theart results on RHAS133 in diverse evaluation settings."

3

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jun 4

Mein Buch “Was kann Künstliche Intelligenz?“ ist erschienen. Eine leicht zugängliche Einführung in das Thema Künstliche Intelligenz. LeserInnen – auch ohne technischen Hintergrund – wird erklärt, was KI eigentlich ist, welche Potenziale sie birgt und welche Auswirkungen sie hat.

3 3

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jun 4

We are soooo proud. Our European-developed TiRex is leading the field—significantly ahead of U.S. competitors like Amazon, Datadog, Salesforce, and Google, as well as Chinese models from companies such as Alibaba.

Michael Kopp @mkk20.bsky.social · Jun 2

Ever though you could get a 35 million parameter time series state-of-the art foundation model that you can run on embedded hardware? Thanks to @hochreitersepp.bsky.social and his team at NXAI, you can. Amazing work!
Paper: arxiv.org/abs/2505.23719
Code: github.com/NX-AI/tirex

TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learning

In-context learning, the ability of large language models to perform tasks using only examples provided in the prompt, has recently been adapted for time series forecasting. This paradigm enables zero...

arxiv.org

4 5

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jun 2

Attention!! Our TiRex time series model, built on xLSTM, is topping all major international leaderboards. A European-developed model is leading the field—significantly ahead of U.S. competitors like Amazon, Datadog, Salesforce, and Google, as well as Chinese models from Alibaba.

Michael Kopp @mkk20.bsky.social · Jun 2

Ever though you could get a 35 million parameter time series state-of-the art foundation model that you can run on embedded hardware? Thanks to @hochreitersepp.bsky.social and his team at NXAI, you can. Amazing work!
Paper: arxiv.org/abs/2505.23719
Code: github.com/NX-AI/tirex

TiRex: Zero-Shot Forecasting Across Long and Short Horizons with Enhanced In-Context Learning

In-context learning, the ability of large language models to perform tasks using only examples provided in the prompt, has recently been adapted for time series forecasting. This paradigm enables zero...

arxiv.org

3 7

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Jun 2

TiRex 🦖 time series xLSTM model ranked #1 on all leaderboards.

➡️ Outperforms models by Amazon, Google, Datadog, Salesforce, Alibaba

➡️ industrial applications

➡️ limited data

➡️ embedded AI and edge devices

➡️ Europe is leading

Code: lnkd.in/eHXb-XwZ
Paper: lnkd.in/e8e7xnri

shorturl.at/jcQeq

Introducing TiRex - xLSTM based time series model | NXAI

TiRex model at the top 🦖 We are proud of TiRex - our first time series model based on #xLSTM technology. Key take aways: 🥇 Ranked #1 on official international leaderboards ➡️ Outperforms models ...

www.linkedin.com

5 5

Reposted

Günter Klambauer @gklambauer.bsky.social · May 30

Recommended read for the weekend: Sepp Hochreiter's book on AI!

Lots of fun anecdotes and easily accessible basics on AI!

www.beneventopublishing.com/ecowing/prod...

1 3 6

hochreitersepp.bsky.social @hochreitersepp.bsky.social · May 26

xLSTM for the classification of assembly tasks: arxiv.org/abs/2505.18012

"xLSTM model demonstrated better generalization capabilities to new operators. The results clearly show that for this type of classification, the xLSTM model offers a slight edge over Transformers."

1 3

Reposted

fses91.bsky.social @fses91.bsky.social · May 22

Happy to introduce 🔥LaM-SLidE🔥!

We show how trajectories of spatial dynamical systems can be modeled in latent space by

--> leveraging IDENTIFIERS.

📚Paper: arxiv.org/abs/2502.12128
💻Code: github.com/ml-jku/LaM-S...
📝Blog: ml-jku.github.io/LaM-SLidE/
1/n

1 8 7

Reposted

Sebastian Sanokowski @sanokows.bsky.social · Apr 24

1/11 Excited to present our latest work "Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics" at #ICLR2025 on Fri 25 Apr at 10 am!
#CombinatorialOptimization #StatisticalPhysics #DiffusionModels

1 7 16

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Apr 24

xLSTM for Multi-label ECG Classification: arxiv.org/abs/2504.16101

"This approach significantly improves ECG classification accuracy, thereby advancing clinical diagnostics and patient care."
Cool.

2

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Apr 10

Precall: 4 Tenure-Tracks in AI for females. Check it out.

Johannes Fürnkranz @juffi-jku.bsky.social · Apr 7

Four Tenure-track positions in AI exclusively for female applicants at @jkulinz.bsky.social! Details available soon at www.jku.at/en/lit-artif...

Precall: Four tenure-track professorships

www.jku.at

4 4

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Mar 19

Huge achievement: arxiv.org/abs/2503.14376
xLSTM kernels are now the fastest kernels both for training and inference. Faster than Flashattention or Mamba kernels. High arithmetic intensity. Optimized use of GPUs that set a new state of the art. Congratulations to the team. It is highly impressive.

arxiv.org

3 8

Reposted

Jonathan Tirone @virtualnomad.bsky.social · Mar 16

Visited with legendary #AI pioneer @hochreitersepp.bsky.social at #JKULinz. We discussed the efficiency of #xLSTM against #GPT and why AI's future may be located on the edge of networks, rather than in centralised data centres via @bloomberg.com www.bloomberg.com/news/article...

AI Pioneer Wants Europe to Forge Its Own Nimbler Way Forward

One belief underlying the power-hungry approach to machine learning advanced by OpenAI and Mistral AI is that an artificial intelligence model must review its entire dataset before spitting out new in...

www.bloomberg.com

3 4

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Mar 14

xLSTM for Automated Stock Trading: arxiv.org/abs/2503.09655

xLSTM outperforms LSTM.
"These findings mark the potential of xLSTM for enhancing DRL-based stock trading systems."

This reinforcement learning approach uses xLSTM in both actor and critic components, which increases the performance.

A Deep Reinforcement Learning Approach to Automated Stock Trading, using xLSTM Networks

Traditional Long Short-Term Memory (LSTM) networks are effective for handling sequential data but have limitations such as gradient vanishing and difficulty in capturing long-term dependencies, which ...

arxiv.org

3 9

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Mar 14

Join Our Research Team in Linz!
We are looking for 5 PostDocs and 10 PhDs in Machine Learning working on xLSTM, NLP, robustness, density ratio.

Deadline: 04/20/25.

More details: www.jku.at/en/lit-artif...

#MachineLearning #DeepLearning #ResearchOpportunities #PhDPositions

Deep Learning

www.jku.at

4 15

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Feb 19

Exploration imitation learning architectures: Transformer, Mamba, xLSTM: arxiv.org/abs/2502.12330
*LIBERO: “xLSTM shows great potential”
*RoboCasa: “xLSTM models, we achieved success rate of 53.6%, compared to 40.0% of BC-Transformer”
*Point Clouds: “xLSTM model achieves a 60.9% success rate”

X-IL: Exploring the Design Space of Imitation Learning Policies

Designing modern imitation learning (IL) policies requires making numerous decisions, including the selection of feature encoding, architecture, policy representation, and more. As the field rapidly a...

arxiv.org

3 6

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Feb 17

xLSTM for time series with Granger causality: arxiv.org/abs/2502.09981

xLSTM again shows superb performance at time series analysis.

"Our experimental evaluations on three datasets demonstrate the overall efficacy of our proposed GC-xLSTM model."

Exploring Neural Granger Causality with xLSTMs: Unveiling Temporal Dependencies in Complex Data

Causality in time series can be difficult to determine, especially in the presence of non-linear dependencies. The concept of Granger causality helps analyze potential relationships between variables,...

arxiv.org

3 15

hochreitersepp.bsky.social @hochreitersepp.bsky.social · Feb 4

xLSTM shines at tumor segmentation: arxiv.org/abs/2502.00314

“evaluated state-of-the-art segmentation methods, including U-Net and its enhanced variants with Transformers and Mamba. Our proposed ViLU-Net [vision xLSTM-Net] model achieved superior performance with reduced complexity.” Cool.

A Study on the Performance of U-Net Modifications in Retroperitoneal Tumor Segmentation

The retroperitoneum hosts a variety of tumors, including rare benign and malignant types, which pose diagnostic and treatment challenges due to their infrequency and proximity to vital structures. Est...

arxiv.org

1 3