Andreas Wierse

"A new study conducted at the University of Vienna, the Max Planck Institute for Intelligent Systems in Stuttgart, and the Helmholtz Centers in Berlin and Dresden takes an important step in the challenge to miniaturize computing devices and to make them more energy-efficient. The work, published in Science Advances, opens up new possibilities for creating reprogrammable magnonic circuits by exciting spin waves by alternating currents and redirecting these waves on demand. The central processing units (CPUs) that we use in our laptops, desktops or even phones use billions of transistors, which are based on the complementary metal oxide semiconductor (CMOS) technology. With the increasing need to miniaturize these devices, several physical limitations are raising concerns about their sustainability. In addition, high power consumption and energy losses push scientists to search for alternative computing architectures." #magnonic #spinwaves

2

With his keynote on “Taming (some) Heterogeneity in the Cloud” Wolfgang Lehner of Technische Universität Dresden contributed to the 2nd CloudDB workshop collocated with the 50th International Conference on Very Large Data Bases in Guangzhou, China VLDB2024.   Data Systems are faced with heterogeneity at different levels – different data models and workload types on the one end and heterogeneous system deployments and HW components on the other end. The talk outlined the challenges and – using the DAPHNE system as an architectural blueprint – discussed ways to overcome heterogeneity in the Cloud. The DAPHNE EU Project is a European Union project aimed at creating an open, extensible infrastructure for integrated data analysis pipelines and allows to cater different technical environments and application scenarios.   The CloudDB workshop also allowed the stage for an in-depth discussion on “Cloud-based Data Management and AI” within a panel run by Viktor Leis, Alvin Cheung, Ji Sun, and Wolfgang Lehner. Different aspects of the impact of AI on the system perspective at the level of compute, storage, and communication as well as on self-*capabilities of large data management systems were presented and discussed.   Thanks for a great VLDB24 in Guangzhou, China! Looking forward to VLDB25 in London, UK!

9

🚀 Exciting News! 🚀 We are thrilled to announce that Kumo AI's position paper "Relational Deep Learning: Graph Representation Learning on Relational Databases," has been accepted at #ICML 2024! 🎉 Join us in Vienna from July 21-27, where our very own Matthias Fey will be presenting this exciting new research area. 🔗 https://icml.cc/ In collaboration with Stanford University, we introduce Relational Deep Learning (#RDL), a new research area for applying deep learning on relational databases by viewing them as temporal, heterogeneous graphs. This method eliminates the need for manual feature engineering, leading to more accurate models built faster. 🚀 We’ve also developed #RelBench, a benchmark suite to facilitate research in this new area, using diverse datasets from Stack Exchange discussions to Amazon book reviews. 📚 Stay tuned for more updates on our presentation format, and check out our paper here: https://lnkd.in/dk24vbSg #ICML2024 #MachineLearning #AI #GraphNeuralNetworks #RelationalDeepLearning #DataScience #KumoAI #StanfordCollaboration

38

Welcome to read the article "Integrated Gradients for Feature Assessment in Point Cloud-Based Data Sets" which is written by Mr. Markus Schwegler, Prof. Christoph Müller and Alexander Reiterer. This excellent article was published in issue 7 in 2023 and has been viewed 1313 times and cited 3 times! Article link: https://lnkd.in/gMCsed5z MDPI Fraunhofer IPM Furtwangen University Albert-Ludwigs-Universität Freiburg (Albert Ludwig University of Freiburg) #pointcloud #neuralnetwork #integratedgradients

2

#DigitalTech 👉 #GreenTransition #HPC In the digital decade, distributed systems, high-performance computing (#HPC) are at the core of major advances and innovation, and a strategic resource for Europe's future. Recent advances in Large Language Models (#LLM), foundation models, and generative AI, such as those addressed in our AI BOOST project, require tremendous computing power to train the models. As a consequence, HPC is indispensable. Computation should also be optimized in an energy-efficient way 👉 #GreenDeal #GreenTech #GreenComputing I had the pleasure to make my contributions in three #bookchapters of the book on Large-scale Distributed Systems and Energy Efficiency: A Holistic View, published by Willey & Sons : Chapter 3 GREEN WIRED NETWORKS 41 Chapter 4 GREEN WIRELESS-ENERGY EFFICIENCY IN WIRELESS NETWORKS 81 Chapter 6 GREEN DATA CENTERS 159 👉 https://lnkd.in/ebWbcmsH. EIT - European Institute of Innovation and Technology European Innovation Council and SMEs Executive Agency (EISMEA) COST Association - European Cooperation in Science and Technology

2

WHAT IS SINGLE PAIR ETHERNET? Single Pair Ethernet (SPE) describes the transmission of Ethernet via only one pair of copper wires In addition to data transmission via Ethernet, SPE also enables simultaneous power supply to end devices via PoDL - Power over Data Line. Traditionally two pairs of copper wires were necessary for Fast Ethernet (100MB) or four pairs of copper wires for Gigabit Ethernet. “Example Cat6A Cable” Now Single-Pair Ethernet “SPE” is represented by a new group of IEEE 802.3 standard can run cable up to 100 meters for 100 Megabit and 40 meters for 1 Gigabits SPE now opens up completely new possibilities and fields of application for industrial Ethernet. #RCDD #Datacenter #TIA942

49

WHAT IS SINGLE PAIR ETHERNET? Single Pair Ethernet (SPE) describes the transmission of Ethernet via only one pair of copper wires In addition to data transmission via Ethernet, SPE also enables simultaneous power supply to end devices via PoDL - Power over Data Line. Traditionally two pairs of copper wires were necessary for Fast Ethernet (100MB) or four pairs of copper wires for Gigabit Ethernet. “Example Cat6A Cable” Now Single-Pair Ethernet “SPE” is represented by a new group of IEEE 802.3 standard can run cable up to 100 meters for 100 Megabit and 40 meters for 1 Gigabits SPE now opens up completely new possibilities and fields of application for industrial Ethernet. #RCDD #Datacenter #IEE

99

1 Kommentar

Supermicro's New Multi-Node Liquid Cooled Architecture with Maximum Performance Density Purpose-Built for HPC at Scale The New Rack-Scale Ready FlexTwin™ Systems Deliver Unprecedented Compute-Density in a Multi-Node Form Factor, with DLC dual CPUs up to 500W Each per Node, Front Node Access and Optimized Networking

17

1 Kommentar

🗣 Had a great time on a high energy panel at the Supercomputing Conference 2024, discussing "Educating for a Hybrid Future: Bridging the Gap between High-Performance and Quantum Computing" 🌉 Honoured to share the stage with Dr. Stefan Seegerer (IQM Quantum Computers), Monica VanDieren (NVIDIA), Bruno Abreu (Pittsburgh Supercomputing Center), and Minh Chung (Leibniz Supercomputing Centre ), and explore the exciting intersection of quantum computing 🖥 and HPC. From discussing the role of humanities students in quantum science 🔭 to the importance of diversity 👦👩 and inclusion 👦👦 in the field, we covered a remarkable spectrum of topics in the short 90 minutes 🕐 We also shared insights on various educational initiatives, workforce development programs, and the fundamental role of math and physics ⚖️ in building a strong foundation for quantum computing and HPC. I was a bundle of nerves before we took the stage. Thanks to the artful moderation by Vincent A. 👏 and the lively dialogue with the audience, I quickly eased into the moment and found myself truly enjoying the session. It was a reminder that stepping outside of your comfort zone can lead to incredible experiences 👣 Looking forward to the next Supercomputing Conference in St Louis, 2025! 🚀 National Research Foundation Singapore National Supercomputing Centre (NSCC) Singapore Institute of High Performance Computing (A*STAR IHPC) National Quantum Computing Hub (NQCH) #marvellouslyyours #sc24 #hpc #quantumtech #quantum #quantumcomputing #tech #research #outreach #education #math #physics #humanities #training #development #nvidia #aws #iqm #quantinuum #lsc #psc #nrf #nscc #workforce #ai

78

3 Kommentare

🔍 Unlocking Quantum Potentials! Recent advancements from Paderborn University highlight the crucial role of high-performance computing (HPC) in the realm of quantum photonics. By utilizing HPC to analyze quantum detector data, researchers are now able to process significant datasets at an unprecedented speed—turning computations that once required hours into mere minutes. ⚡ 💡 This game-changing approach, pioneered by physicist Timon Schapeler and Dr. Robert Schade, is laying the groundwork for effective characterization of photon detectors, which are essential for quantum measurement applications. The implications are significant, not just for system characterization but also for the evolving landscape of photonic quantum computers, pushing the boundaries toward achieving quantum supremacy. 🌌 🔧 As the field of quantum states of light continues to expand, professionals in software development, marketing, AI, and SaaS must stay aware of these transformative developments. Engaging with high-resolution photonics can lead to groundbreaking innovations in measurement technologies, data processing, and communications. What are your thoughts on the integration of HPC in quantum research? 🤔 ✨ Stay Ahead in Tech! Connect with me for cutting-edge insights and knowledge sharing! Want to make your URL shorter and more trackable? Try linksgpt.com #QuantumResearch #Photonics #BitIgniter #LinksGPT Want to know more: https://lnkd.in/e3XaD_hZ

2

Graph Mining Breakthrough ( Advancing Network Analysis with Triangle-Dense Subgraphs) Follow me for more technical insights: https://lnkd.in/evN3jc94   The field of graph mining has taken a significant step forward with the development of a new computational algorithm led by Nikolaos Sidiropoulos from the University of Virginia and Aritra Konar from KU Leuven. This breakthrough tackles one of the most challenging tasks in network analysis—locating triangle-dense subgraphs, an essential problem for applications in fraud detection, computational biology, and data analysis.  >>> Key Technical Insights:  - Triangle-Dense Subgraphs:  Traditional graph mining algorithms focus on finding connections between pairs of nodes. This new method expands on this concept by examining triangles—groups of three nodes, each connected to the others. This approach is critical for uncovering tightly connected groups, such as friend clusters in social networks or interacting genes in biological systems. - Submodular Relaxation for Efficiency:  The algorithm tackles the computational challenges of identifying triangle-dense subgraphs by employing submodular relaxation. This technique simplifies the complex problem, allowing it to be solved more quickly without sacrificing the precision needed for large, intricate networks. - Applications in Diverse Domains:  - Fraud Detection: Identifying suspicious patterns in financial networks.   - Social Media: Discovering community dynamics among highly interactive users.   - Computational Biology: Mapping complex protein interactions and genetic relationships with improved precision.  >>> Published Work:  This research has been published in IEEE Transactions on Knowledge and Data Engineering, reflecting the importance of these advances for professionals and researchers working with large datasets. As datasets grow in size and complexity, the ability to identify meaningful patterns across multi-connection relationships will become increasingly vital. This innovative algorithm marks a new chapter in the evolution of network analysis tools, enhancing capabilities for fraud prevention, social network analysis, and scientific discovery. https://lnkd.in/dUzJhnV9 #GraphMining #DataAnalysis #NetworkScience #FraudDetection #SocialMediaAnalytics #ComputationalBiology #AIResearch #IEEE

7

🔍 Don't Miss Out on This Insightful Session! 🔍 Join us at Data Week Leipzig for a dynamic presentation on ""Industrial Knowledge Graphs in action"" by Nikolay Krustev from eccenca's partner Ontotext 🕒 When: April 18th, between 13:00 - 14:30 📍 Where: Sitzungssaal 🎙️ Language: English Nikolay Krustev, leading Ontotext's Industrial Vertical, brings over a decade of experience in driving strategic business development across Automotive, Manufacturing, AECO, and Energy sectors. In this session, Nikolay will unveil real-world applications and use cases, demonstrating the power of knowledge graphs in transforming industries. From enhancing data product catalogs to architecting modern data stacks for commercial analytics, Nikolay's insights promise to inspire and enlighten. Don't miss this opportunity to delve into the world of industrial knowledge graphs and gain valuable insights from Nikolay and other industry experts. See you there! #KnowledgeGraphs #DataWeekLeipzig #IndustrialInsights #eccenca #Ontotext

23

Fraunhofer's OpenGPT-X project releases Teuken-7B on Hugging Face ✅ 7 billion parameters ✅ Trained in all 24 EU languages ✅ 50% non-English data (compared to 8% in LLaMA-3) ✅ Major languages involved: English, French, German, Italian, Spanish. ✅ Context Window: 4K ✅ Open-source & commercially usable ✅ Optimized for multilingual efficiency and cost reduction 🔹 Ideal for European organizations with multilingual needs 🔹 Ensure sensitive data stays secure within your infrastructure 🔹 Customize it for industry-specific applications (e.g., finance, medicine) I can't wait to try it on #Ollama!! Source: https://lnkd.in/d8JbdXc7 Paper: https://lnkd.in/dS5CQZRU Tips: https://lnkd.in/dyQpnSEJ #GenAI #Teuken7B #OpenGPTX #MultilingualAI #Fraunhofer #HuggingFace #OpenSource #EuropeanUnion #EU #TechInnovation #DataSovereignty

14

🌟 Another SwissText Keynote revealed 🌟 We are thrilled to invite you to an insightful keynote titled "Attacks on large language models" by Leon Derczynski, affiliated with NVIDIA and ITU Copenhagen, at the Swiss Text Analytics Conference in Chur.     🔍 Abstract: Generative language models got a lot better recently; they escaped the research "yard" and now act in public. This has the consequence of risking reputational and other harms as models produce undesirable output. Simultaneously, generative language models have become alluring targets for hackers, who attempt to subvert them. Unlike traditional information security, the entry-level language for interacting with language models is natural language (at least for English), and so no coding skills are required. This has led to a huge effervescent and creative frontier where thousands of people attack language models in an attempt to make them misbehave. We investigate how people attack language models - what motivates them, what their techniques and strategies are, how to describe the activity, and how attacks are developed. And, of course, what the attacks are. New resources have emerged for automatically attacking language models and for collecting various attacks from the community, forming an active baseline for discovering language model vulnerabilities. We will discuss the issues involved in collating attacks on language models, and how to use this information to make model interaction safer.     🗓️ Date: 11.06.2024 🕒 Time: 16:30 - 17:30 📍 Location: Swiss Text Analytics Conference, Chur, FHGR Fachhochschule Graubünden     Don't miss this opportunity to gain valuable insights from one of the leading experts in the field. Join us to explore innovative ideas and engage in thought-provoking discussions. #Keynote #SwissTextAnalyticsConference #Event #Innovation #Networking #ThoughtLeadership #Learning #IndustryInsights Corsin Capol, Albert Weichselbraun, Mark Cieliebak, Manuela Hürlimann, Claudiu Musat, Elisabeth Maier

5

Given the phenomenal amount of data generated during the chip manufacturing process of the 𝐬𝐞𝐦𝐢𝐜𝐨𝐧𝐝𝐮𝐭𝐨𝐫 𝐢𝐧𝐝𝐮𝐬𝐭𝐫𝐲, it is vital that access to the data is both easy and immediate. The use of 𝐩𝐲𝐭𝐡𝐨𝐧 has been democratised in this sector, particularly for access to databases but also to specific files such as 𝐒𝐓𝐃𝐅 𝐟𝐢𝐥𝐞𝐬.  Version 18 of 𝐉𝐌𝐏 offers direct integration of Python to facilitate and accelerate these accesses. This means that 🔸 RAM is directly shared between Python and JMP,  🔸 The JMP environment now includes a Python editor. ☑ Python experts in the semiconductor world can take advantage of JMP's interactive reporting capabilities, without leaving their area of expertise and losing data management capacity. ☑ They can provide JMP applications, potentially incorporating data access in Python, that are easy to use and get straight to the point of understanding the relationships between data.  ☑ Teams can devote even 𝐦𝐨𝐫𝐞 𝐭𝐢𝐦𝐞 𝐭𝐨 𝐢𝐦𝐩𝐫𝐨𝐯𝐞𝐦𝐞𝐧𝐭 𝐚𝐧𝐝 𝐢𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧! For those curious, this discussion led by our development expert may interest you https://lnkd.in/eDvUX8Ck

28

1 Kommentar

#Optical computing promises to improve the speed and energy efficiency of #machinelearning applications. However, current approaches to efficiently train these models are limited by in silico emulation on digital computers. Here Chinese Researchers have developed a method called fully forward mode ( #FFM ) learning, which implements the compute-intensive training process on the physical system. The majority of the machine learning operations are thus efficiently conducted in parallel on site, alleviating numerical modelling constraints. China is innovating at a faster clip!. The timing of Taichi-II’s debut is particularly notable. As the US has imposed restrictions on China’s access to advanced #GPUs for #AI training, the Taichi-II chip offers a viable alternative that could help mitigate these limitations. Additionally, the performance of Taichi-II comes amid reports that NVIDIA’s high-tech AI chips may be making their way into the hands of Chinese military officials, potentially influencing China’s technological advancements.

4

Releasing Llama 3.1 as open source carries significant implications. By open-sourcing a 405 billion parameter model, Zuck claims to democratise advanced AI, enabling researchers, developers, and companies to innovate without prohibitive costs. He intends to encourage global collaboration and rapid advancements as the community builds on Meta’s work. Zuck argues the economic impact will be substantial, with startups and small businesses leveraging state-of-the-art AI capabilities, potentially driving new business models, products, and services. He also advocates for educational institutions, gaining resources to enhance AI education and prepare a skilled workforce. I’m with him on the former, but I’m not so sure about the latter. This is clearly a bold move from a company with a business model that doesn’t rely on providing AI as a service for revenue. Lucky him. He’s firm in his belief that open source will raise all boats on a rising tide. We’ll see. Hope he’s right. Many will claim this openness comes with ethical and safety concerns—deepfakes and all that. I don’t think there’s any more risk than with closed-source models because closed source simply means not yet visible to outsiders, including bad actors. Really? How do we know that? In any case, it won’t stay this way forever. One way or another, it will be exposed. So having billions of eyeballs working on and learning how to use and misuse a major model is better than having this done by who knows whom behind closed doors. Sunlight is the best disinfectant. It will be interesting to see if Meta’s strategy sets a new benchmark in the competitive AI landscape, prompting other companies to adopt similar open-source approaches. One intriguing aspect is how Llama 3.1 promises to open new research avenues, especially in synthetic data generation and model distillation. Zuck talks about everyone having “their own model.” It sounds like a Tower of Babel, but it does mean that one can take a very powerful model as a baseline, and tweak it for specific purposes, whether that’s improved customer service or orchestrating the takeover of Taiwan with autonomous weapons.

6

4 Kommentare