Artificial Intelligence Institute of South Carolina

Attending #EMNLP at Miami 11-16 Last year, we introduced the AI Detectability Index (ADI). This year, we extend the idea to a new language, Hindi. ???? ?????????????????????????? ?????? ????????? - https://lnkd.in/es4Ysue6 Moreover, we empirically examine the limits of text watermarking vs. distortion Looking forward to meeting #NLPProc colleagues, friends..... Ishan Kavathekar, Anku R., Ponnurangam Kumaraguru (PK), Ashmit Chamoli, Amit Sheth

?? Excited for an incredible week exploring causality and neuro-symbolic AI at #ISWC2024 in Baltimore! If you're passionate about advancements in #CausalAI and #NeuroSymbolicAI, here’s a glimpse into what we’re bringing to the table: 1?? **#RelationType**: Join us on Day 1 at the WOP workshop, where we’ll present our paper on *Ontology Design Metapattern for RelationType Role Composition* (Jaimini et al., 2024). We’re delving into ontology design patterns that push the boundaries of relation modeling! Read more here: https://lnkd.in/e7Fs3hdS 2?? **#VisualCausalQA**: On Day 3, catch our poster and demo on *Visual Causal Question and Answering with Knowledge Graph Link Prediction*. We’re leveraging causal neuro-symbolic AI to enhance scene understanding with causal Q&A. Check out the work here: https://lnkd.in/eujVSpA2 3?? **#CausalKG**: Day 4 on the industry track, we present *Causal Knowledge Graph for Scene Understanding in Autonomous Driving*. Discover how causal insights can reshape autonomous driving safety and reliability. Read more here: https://lnkd.in/e5jeHmh2 4?? **#CausalNeurosymbolicAI**: Wrapping up on Day 5, we dive into *Causal Neuro-Symbolic AI for Root Cause Analysis in Smart Manufacturing*. Join us to see how causality-driven AI is transforming diagnostics in manufacturing. More details here: https://lnkd.in/enbFA6fA Let’s connect! Whether you’re attending or just curious, drop a DM, and let’s discuss the future of #Causality, #NeuroSymbolicAI, and #ExplainableAI. Looking forward to pushing the boundaries together!? #ISWC2024 #Causality #NeurosymbolicAI #ExplainableAI #AutonomousDriving #RootCause #SmartManufacturing Cory Henson Ruwan Wickramarachchi Amit Sheth

Our Generative AI in Health Hackathon is a one-week take-home event kicking off on November 9 and concluding on November 16-17, co-hosted by NVIDIA, StartX, and HealthUnity. Why You Should Participate: - Address real-world challenges in Generative AI for healthcare. - Build LLM agents using state-of-the-art software and health datasets. - Expand your network and connect with experts in the field. These are only a few highlights! Throughout the one-week event, we offer support with LLM agent frameworks, particularly #openCHA, and provide a range of software and datasets on mental health, nutrition, to mention a few. A special thanks to our sponsors, including Define Ventures, IEEE EMBS, and iHealth Labs, and collaborators for providing the tools for this hackathon, including NVIDIA, StartX, UCI-Institute for Future Health, Artificial Intelligence Institute of South Carolina, Turun yliopisto - University of Turku. Find more details about the tools and datasets here: https://lnkd.in/gUhQ2NiU https://lnkd.in/gZBZh2Qg Register today! https://lu.ma/d0mc06fc Don't miss this opportunity to showcase your skills and make a tangible impact in AI and healthcare! Amit Sheth, Revathy Venkataramanan, Kanak Raj, Jennifer Auxier, Yuning Wang, Mohammad Feli, Kianoosh Kazemi, Anna Axelin, Pasi Liljeberg Jia Li, Jun (Alex) Gao, Carmen Wiedenhoeft, Sara Newsome, Mona Mahmoudi, Ramesh C Jain, Matthew Liu, Paula Wall, Mahyar Abbasian, Amir M. Rahmani, Avneet Kaur, Renu Bhatia, Renee Y., Jack Resnick, Chelsea Sumner, PharmD, Rph, Julia Levites, Brad Genereaux, Mitesh Patel, Darsh Singh Mann, Michael Niaki, Pengpeng Wang, Zejia Shao #LLM, #Hackathon #Agents, #openCHA #HealthUnity2024

Gartner: By 2027, over 50% of GenAI models used by enterprises will be industry or business function-specific (up from ~1% in 2023). Lori Perri Daron Acemoglu (Nobel in Economy, 2024): AI can currently do only 5% of the Jobs. So, what types of models are needed if currently popular?#GenAI?models cannot meet the current and near-future needs of mission-critical and Enterprise AI applications? Answer: Models that are #Trustworthy and #Robust. How will you make these models? By making them #Targeted or #custom, #compact or #right_sized (hence cloud, fog, edge deployable), #neurosymbolic (not just #quality_data but also #knowledge and #human_expertise), and #on_demand (retrain and deploy often as data or knowledge change, with much smaller effort than today). Using real-world #healthcare examples (do check the #videos), my keynote, "Why do we need custom or targeted, compact, and neurosymbolic AI models for health applications?" addresses the above questions. https://lnkd.in/eAjUzuPA May consider checking the abstract (link from cover slide). Thanks,?Professor Dhaval Thakker,?for the invitation. Additional details are in my KiL-KDD keynote a little over a month ago: Forging Trust in Tomorrow’s AI:? A Roadmap for Reliable, Explainable, and Safe NeuroSymbolic AI Systems https://bit.ly/KDD-KiL Interested in Neurosymbolic AI at Artificial Intelligence Institute of South Carolina? https://lnkd.in/eGkZzNVJ

????? I'm excited to announce that our survey paper, "?? ?????????????????????????? ???????????? ???? ???????? ?????????? ????????????????????", has been officially published in ?????? ?????????????????? ??????????????, [???????????? ????????????: ????.??; ????????: ??/?????? (????)]! https://lnkd.in/gGhcjtuw ???????????? ???? ???????? ?????????????????? ???????????????????? ?????????????????? ???? ????????????????: https://lnkd.in/g-YKUM3e ?? Consider catastrophic machinery breakdowns, natural disasters, or even stock market crashes – these rare events might not happen often, but they can have a massive impact.??.?Our paper dives deep into the latest approaches for predicting these infrequent yet tricky occurrences, breaking it down into four key parts: the data, processing techniques, algorithms used, and how to evaluate the results. ???? We have introduced the concept of “???????????? ???? ????????????,” categorizing rare events into four levels: Extremely-rare, Very-rare, Moderately-rare, and Frequently-rare ???. Our study spans 73 rare-event datasets across multiple industries, encompassing diverse modalities such as numerical, image, text, and audio. ????????We explored four data processing methods, five algorithmic strategies, and two evaluation approaches. ?????? ???????????????????? ???????? ???????????? ???? ??????????! Key challenges include the need for advanced data processing techniques to address extreme data imbalance, the lack of standardized benchmark datasets for consistent evaluation, and the limited development of evaluation methods specifically designed for rare events. ??Looking ahead, we foresee significant opportunities in rare event prediction, particularly in areas like explainability, process knowledge workflows, and automated planning strategies to mitigate rare events. ??? Thanks to Ruwan Wickramarachchi and Dr. Amit Sheth for their incredible support! Feel free to reach out if you’d like to discuss the details or share your insights!?? Visit here for more on our smart manufacturing project: https://lnkd.in/gCgcQT4C #smartmanufacturing #rareeventprediction #timeseries #anomalydetection #forecasting #Industry40

If you are curious about what academic life, I highly encourage you to listen to my conversation with Prof Amit Sheth, from University of South Carolina, where we talk about: + How to lead an extremely successful #research lab + The secrets to being successful in #academia + Finding, recruiting and mentoring excellent #GraduateStudents + Reflecting on a #lifetime of achievements Here is a brief overview: https://lnkd.in/gyyvUQY8 You can watch the complete videos here: https://lnkd.in/g7YhQ6CV https://lnkd.in/gQAbbXW3 https://lnkd.in/g4RT_Nh7 https://lnkd.in/g3hZr7fd If you want me to ask other questions in later sessions from other colleagues and friends, feel free to put them in the comments! #phdhelp #phdlife #phdjourney #phdstudent #phd #gradschoolproblems #roadtophd #phdproblems #phdcandidate #graduateschool #thesisproblems #phdstudentlife #dissertationproblems #doctorallife #academiclife #gradschool #gradstudentproblems #doctoralstudent #gradstudent #gradschoollife #phdwriting #academia #masters #mastersdegree #college #thesisproblems #dissertationdone #university