Centro ALGORITMI’s Post

Exploring Parameter-Efficient Fine-Tuning (PEFT): LoRA & QLoRA PEFT revolutionizes model fine-tuning by minimizing computational resources and costs while enabling effective domain-specific adaptations. Among the most impactful techniques are Low-Rank Adaptation (LoRA) and its advanced variant, Quantized Low-Rank Adaptation (QLoRA). 🔍 LoRA: Modular low-rank networks enable memory-efficient fine-tuning without altering the original model weights, reducing memory footprint significantly. 🔍 QLoRA: Takes it a step further with 4-bit quantization for faster calculations and reduced memory demands, making it ideal for cost-effective optimization. These methods allow us to achieve modularity, avoid catastrophic forgetting, and enhance computational efficiency for domain-specific tasks. 🚀 📖 Dive deeper into these techniques in this insightful blog: https://lnkd.in/gfbzSAP2 #GenerativeAI #LoRA #QLoRA #PEFT #AIInnovation #MachineLearning #DeepLearning #ModelOptimization

⚡ #AI4AM2024: 𝐀𝐫𝐭𝐢𝐟𝐢𝐜𝐢𝐚𝐥 𝐈𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞 for Advanced Materials 🔎 A cross-disciplinary international event bringing together top experts from #industry and #research using #AI for #materialsscience and #engineering. 🗣️ Several partners of MaX Centre for Materials at the exascale contributed to the conference by chairing sessions and delivering talks. 🤝 MaX Centre for Materials at the exascale also supported the event as a sponsor, together with international #tech and R&D companies. 🔍 Delve into the details of the conference here: https://bit.ly/3W6mxhT #materials #supercomputing #CoE

As OntoTrans H2020 approaches to its closing stages, it is a good time to reflect on over four years of intensive collaborative research in the project. Check out this press release to find out more! #interoperability #data #SemanticMaterials #Innovation #Research #HorizonEurope #Horizon2020 #KnowledgeManagement

🚀 Excited to share that our new paper has been published in Transportation Research Procedia! 📄 Title: A multi-objective two-echelon vehicle routing problem with multiple delivery options 🔍 Summary: We've explored a new approach to the two-echelon vehicle routing problem (2E-VRP) where both echelons can make home deliveries. Our sustainable model considers customer preferences, economic, and environmental costs. We also developed an Adaptive Large Neighbourhood Search (ALNS) meta-heuristic algorithm that effectively solves small and large real-world instances. 💡 Key Points: - Dual-echelon vehicle used for deliveries. - Focus on sustainability and customer preferences. - Efficient algorithm for real-world applications. I would like to express my heartfelt gratitude to my supervisors, Dr. Lele Zhang and Dr. Mark Fackrell, as well as, Prof. Russell Thompson, for their invaluable support and collaboration. Thank you all for your guidance and teamwork. Read the full paper here: https://lnkd.in/g-HHb6Sf #Logistics #VehicleRouting #Sustainability #Transportation #SupplyChain #Optimization #DataScience #MetaHeuristic #TransportationResearchProcedia

🎇 Breaking news 🎇 RFF Trøndelag (Regional Research Fund Trøndelag) will support Nicoustic by funding a Regional Qualification Project. The topic of the project is «Use of AI to interpret acoustic sensor data». The project will be carried out in collaboration with SINTEF Digital in 2025. Read more on our website 👇

Article “THE ASSESSMENT OF THE IMPACT OF ARTIFICIAL INTELLIGENCE ON TRANSPORTATION SUSTAINABILITY” by Elena Kaledinova and Dianthy Boeldak We are proud to announce that our article has been published in Scientific Journal of Silesian University of Technology. Series Transport. URL (.pdf, open access): https://lnkd.in/etwrvvNr Journal, volume 125, 2024: https://lnkd.in/eSyE_8qN #artificialintelligence #sustainability #transportationsustainability #logistics #logisticssustainability #sustainableperformance #appliedresearch #researcharticle #sustainabilityassessment

I am delighted to share that our research paper titled: "#Optimizing Green #Supply Chain: A Hub Location Model Utilizing #NSGAII and #MOPSO Algorithms for Enhanced #BTS Site #Coverage in #Telecommunication Networks" has been accepted for publication in the prestigious #Algorithms journal #MDPI (Q1 JCR;Impact Factor: 1.8, CiteScore: 4.1). This achievement would not have been possible without the invaluable support and guidance of my esteemed PhD supervisors: Dr. Mehran Khalaj, Dr. Ramin Enayati, and Dr. Mehdi Keramatpour. I extend my heartfelt gratitude to them for their mentorship, which played a pivotal role in this success. Abstract: Today, #facility location #planning primarily pertains to the long-term strategic and operational #decision-making of large public and private organizations, and the significant costs associated with facility location, construction, and operation have turned location research into long-term decision-making. Presenting a #hub location model for the green supply chain can address the current status of facilities and significantly improve demand #coverage at an acceptable cost. Therefore, in this study, a network of facilities for hub location in the service site domain, considering existing and potential facilities under probable #scenarios, has been proposed. After presenting the #mathematical model, validation was performed on a small scale, followed by sensitivity analysis of the main parameters of the model. Furthermore, a #metaheuristic algorithm was employed to analyze the #NP-Hardness of the model. Additionally, two metaheuristic algorithms, #NSGAII and #MOPSO, were developed to demonstrate the efficiency of the model. Based on the conducted analysis, it can be observed that the computational time increases exponentially with the size of sample problems, indicating the NP-Hardness of the problem. However, the NSGAII algorithm performs better in terms of computational time for #medium-sized problems compared to the MOPSO algorithm. These algorithms were chosen due to their proven efficiency in handling NP-hard #optimization problems and their ability to balance #exploration and #exploitation in search spaces. Mehdi Keramatpour Ramin Enayati Mehran Khalaj https://lnkd.in/dA9bgXsV

Nicoustic receives Qualification Project funding from RFF Trøndelag! This is great, with this support we can look into an exciting but risky venue - using AI for acoustic signal processing.

We are moving forward with our #digitalization and #AI project applied to #biostratigraphy with the Faculdade de Ciências e Tecnologia da Universidade NOVA de Lisboa as partners in an MSc project. We are supported by the AIBC EUROCLUSTERS Project #AIBCforSMEs of the European Innovation Council and SMEs Executive Agency (EISMEA) and European Cluster Collaboration Platform (ECCP).

Master the prompt Engineering technique to get high quality answers. #GenerativeAI #AIPrompts #ArtificialIntelligence #MachineLearning #AITechnology #PromptEngineering #AIInnovation #TechTrends #FutureOfAI #AIApplications #AIwithSatheesh