TetraScience’s Post

Science recently published a story about a researcher that had a postdoc in his lab fabricating data.   In this instance, they were fortunate that someone identified the fabrication before the papr was published.   Proofig AI is a valuable tool for Principal Investigators and lab instructors, offering a streamlined approach to validate the quality of their scientists’ work and ensuring its readiness for publication.   Read the full article here and let us know your thoughts: https://bit.ly/3Si3qzz   #ResearchIntegrity #AcademicIntegrity 

🚀 Exciting times in healthcare! My latest blog explores how AI is transforming the industry, from drug discovery to patient care. Discover the potential of machine learning, deep learning, and natural language processing in enhancing outcomes and operational efficiency. A big thank you to my co-authors: Shrey, Tripti, and Tanishq from CSAI-B for their invaluable contributions! Let’s continue to innovate and push the boundaries of what’s possible in healthcare! 💡🌟 #AI #HealthcareInnovation #MachineLearning

Exciting developments in AI for scientific research! 🚀🧠 SciAgents: A groundbreaking approach to autonomous scientific exploration Key features: -Large-scale ontological knowledge graphs -Suite of LLMs and data retrieval tools -Multi-agent systems with in-situ learning SciAgents aims to tackle a major AI challenge: creating systems that can autonomously advance scientific understanding by: -Exploring novel domains -Identifying complex patterns -Uncovering hidden connections in vast scientific datasets This innovative approach could revolutionize how we conduct research and accelerate scientific discoveries across disciplines. #ArtificialIntelligence #ScientificResearch #MachineLearning #DataScience #Innovation #TechAdvancement #AIforScience #KnowledgeGraphs #MultiAgentSystems https://lnkd.in/dvEHaQp8

A Framework for Extending Co-creative Communication Models to Sustainability Research - Frontiers: The outcomes of computational creativity research based on AI technologies, such as distributed artificial intelligence and self-organizing networks, ... http://dlvr.it/TBVv3D

The advent of FDA-cleared artificial intelligence tools in clinical settings marks a significant milestone in bringing these technologies from research labs into clinical practice. What opportunities and challenges lie ahead? Read more in the AI Assistant blog: https://bit.ly/45nau2E

Thanks to OECD for organizing this panel on AI regulation, and to PILAR DEL CASTILLO and all other panelists for a very insightful discussion on how to ensure competition. In my view, this can be achieved by fostering open innovation, thought transparency mechanisms and data/compute access. #airegulation #artificialintelligence

With the current state of LLM's The only way this can yield something worthwhile is if one has access to a ton of compute and ton of researcher-man-hours to validate the results. As of now LLM's are not the right tools for generating "useful" never-seen-before ideas. It may be able to create something that looks innovative to us as we may not have thought about it but it might exist somewhere in its training data, inside some arcane document that we never read about. Even if it does come up with a novel idea it doesn't have the capability to reason and validate (unless even that is documented somewhere).

Exciting news! 🎉 Our HyperPCM model has been published in the latest special edition of JCIM, Machine Learning in Bio-Cheminformatics. 🧬HyperPCM is the first HyperNetwork approach dedicated to learning drug-target interactions with enhanced generalization to zero-shot tasks. What does this mean for drug discovery? HyperPCM transfers knowledge between protein targets, enabling the prediction of interactions between potential drugs and targets without prior data. The lowered computational costs of the produced target-specific models also facilitate large-scale virtual screening, essential for the rapid development of drugs. Not yet convinced? HyperPCM achieves state-of-the-art performance on several benchmarks by utilizing advanced, pre-trained encoders of molecular structures and protein sequences, and by enriching these embeddings with information from a greater context using a Modern Hopfield Network. Dive into the capabilities of HyperPCM on Hugging Face 🤗: https://lnkd.in/dcZwvBm8 Check out our paper and code: 📰 https://lnkd.in/dg6NRZdu 👩💻 https://lnkd.in/dRz5Z9Gs HyperPCM is joint work with Pieter-Jan H., Sepp Hochreiter, and Günter Klambauer as part of AIDD-ITN. Thank you for your invaluable contributions and supervision! #DrugDiscovery #HyperNetworks #DeepLearning #MachineLearning #AI

Self-Evolution of LLMs Provides a comprehensive survey on self-evolution approaches in LLMs. I often get asked what's next for LLMs? I think it might be self-evolution or something closely related to this emerging paradigm. There is still a lot to solve if you look closely at the different phases such as experience acquisition and self-evaluation. Paper: https://lnkd.in/euPJGjEt --- Tracked by 50K+ researchers and developers, you can also follow my weekly summary of the top AI and LLM papers here: https://lnkd.in/e6ajg945

The webinar titled "The Many Roles of Computational Assistance and AI in the Discovery of New Molecules" focuses on how computational methods and artificial intelligence are revolutionizing the discovery and development of new molecules. The event covers various aspects, including the integration of AI in research processes, how it enhances predictive capabilities, and its role in accelerating innovation in materials science and chemistry. For more detailed information, you can view the complete webinar video recording under the following link: https://lnkd.in/ek_2Mn-8