Francesca Lazzeri, Ph.D.

Researchers at CDS have unveiled that transformer language models can solve complex problems without clear intermediate reasoning, challenging existing assumptions about their operation. Led by CDS PhD students Jacob Pfau and William Merrill, and CDS Associate Professor of Linguistics and Data Science Sam Bowman, this study highlights the use of nonsensical filler tokens that achieve perfect accuracy in specific tasks, raising concerns about  interpretability and accountability in LLMs. Explore the full findings: https://lnkd.in/eERzzKsU

13

Professor Eugene Santos co-authored "Bayesian-knowledge driven ontologies: A framework for fusion of semantic knowledge under uncertainty and incompleteness" published in PLOS ONE. The paper describes how to fuse multiple conflicting ontologies into a single knowledge base. "Biomedicine's rapid advancement is inundating us with new words, labels, and concepts that can be duplicative or even contradictory," says Santos. "Ontologies are necessary to meaningfully link the terms and concepts together. Our framework allows for fragments of any number of ontologies to be combined to create a new ontology that suits the desired domain, regardless of contradictions." Read more: https://lnkd.in/e4bef4jc

31

1 Comment

Fine-tuning LLMs with Reinforcement Learning from Human Feedback This blog post explores how Reinforcement Learning from Human Feedback (RLHF) can be used to fine-tune Large Language Models (LLMs) to align with human preferences. It delves into the concept of Proximal Policy Optimization (PPO) and how it is used in this context. The key takeaway is that RLHF offers a powerful framework to train LLMs that can produce human-aligned text, while still maintaining creativity. This paves the way for more natural and nuanced language generation in the future of AI and natural language understanding. Read the article: https://lnkd.in/d9EGf579 #AI #machinelearning #goML

8

📕Call for Reading: This #specialissue "Surveys in Algorithm Analysis and Complexity Theory" published 5 high-quality papers, and is viewed by 15955! 👉Thank the Guest Editor Dr. Jesper Jansson's efforts on this Special Issue! 🔗We invite you to read and share: https://lnkd.in/gytzk-tY MDPI Kyoto University #algorithmanalysis #complexity #approximationalgorithm

1

This week in #TheBatch: 🎨 All about Claude’s Artifacts feature 🌱 AI growth hinders carbon emissions goals 🧠 GaLore, a memory-efficient method for fine-tuning and pretraining Plus: Andrew Ng shares his concerns about the proposed California regulation SB 1047. Read The Batch now: https://hubs.la/Q02G5YpM0

49

Introducing the Scientific Generative Agent (SGA) framework, a revolutionary unified machine learning approach combining large language models (LLMs) and simulations. Developed by researchers from MIT CSAIL, CMU LTI, UMass Amherst, and the MIT-IBM Watson AI Lab, SGA is transforming cross-disciplinary scientific discovery. Learn how SGA enhances hypothesis generation, experimental design, and data analysis, driving innovation across various scientific domains. Dive into the details and explore its potential implications. Read more - https://lnkd.in/gvpjcmJA 😍 Follow AI Toolhouse for more such amazing content. 🌟 Explore 𝟑𝟔𝟎𝟎+ latest AI Tools here for FREE ➡️ https://lnkd.in/dpQB7xZU #AI #MachineLearning #ScientificDiscovery #SGAFramework #Innovation #Research

1

Dive into the future of systematic literature reviews with our latest blog: "Transforming SLRs: Leveraging AI for Efficiency and Rigor" by Deepika Thakur, Tim Disher, and Imtiaz Samjoo. Discover how AI revolutionizes each step, ensuring efficiency and quality. Read more: https://ow.ly/bFJH50RvQhu. Looking to learn more? Meet our experts at #ISPOR to learn more about AI-powered SLRs. Schedule a meeting: https://ow.ly/13OS50RvQEX. #AI #SystematicReviews

52

Have even more questions about ontologies and Open Knowledge Networks (OKNs)? Read our interview with Dr. Srividya Bansal! How can these frameworks be leveraged to understand complex relationships between data? Learn more in our profile with Dr. Bansal or by browsing the NSF Proto-OKN website: https://www.proto-okn.net/ National Science Foundation (NSF) #science #data #datascience #education #ontology #STEM #computerscience #AI #ML #engineering

6

Wall Street Scholars thrilled to share the latest research from Professor Rama CONT on the Asymptotic Analysis of Deep Residual Networks. This groundbreaking work dives deep into the mathematical foundations of deep learning, shedding light on how residual networks function and perform at scale. Check out the full research and explore the innovative findings that could reshape our understanding of AI and machine learning. https://lnkd.in/e_dHnppJ #DeepLearning #AI #MachineLearning #Research #Innovation #AsymptoticAnalysis #DeepResidualNetworks #WallStreetScholars

3

Advancing Ethical AI: Preference Matching Reinforcement Learning from Human Feedback RLHF for Aligning LLMs with Human Preferences Quick read: https://lnkd.in/gr-238BN Paper: https://lnkd.in/gJddxRws

15

In this week's blog post, Data Science Fellow Sharon H. Green, PhD, MPH writes about how to create powerful visualizations for propensity score matching in R. If you're new to causal inference and trying to learn what tools are available to you, this is a great blog post to check out! Subscribe to our Medium for weekly blog posts on all topics data science, social science, and social justice. https://lnkd.in/epgPtduB

4

DEADLINE APPROACHING! 🌟 Are you headed to EDM 2024 in Atlanta? Consider submitting to the workshop on Causal Inference in Educational Data Mining (EDM) research! Join SEERNet members Neil Heffernan and Avery Closser as they explore the critical role of causality in enhancing student learning through intelligent tutoring systems, educational games, MOOCs, and more. In today's digital learning landscape, understanding causality goes beyond predicting learner behaviors to unraveling the underlying factors impacting educational outcomes. The workshop aims to stimulate discussion and share insights on examining causality within EDM contexts and leveraging EDM to support causal inference methodologies. 📝 Call for Papers/Presentations: To participate, please submit on topics including A/B testing, Bayesian networks, analyzing data from randomized experiments, multi-armed bandits, causal mechanism analysis, and more. Whether you're presenting completed work or exploring open problems in the field, you are invited to contribute to the conversation! - Submission Deadline: June 10, 2024 [Link: https://bit.ly/3WfbSC2]

2

Utah’s leap into data science education is a model for the nation. Read about their success establishing data science as "common ground" between cross-sector stakeholders and their future plans in this insightful Mathematical Association of America article: https://ow.ly/iq7p50SKfQm #DataScience #UtahEducation

4

New paper discusses the role of AI in academia. Aleksei Turobov, PhD Diane Coyle & Verity Harding advocate for exploring & integrating AI tools such as ChatGPT to enhance qualitative research while maintaining the depth of human analysis. They discuss the use of the GPT model for initial coding in qualitative thematic analysis using a sample of United Nations (UN) policy documents. The primary aim of the study is to contribute to the methodological discussion regarding the integration of AI tools, offering a practical guide to validation for using GPT as a collaborative research assistant. The authors outline the advantages and limitations of this methodology and suggest strategies to mitigate risks. Emphasising the importance of transparency and reliability in employing GPT within research methodologies, they argue for a balanced use of AI in supported thematic analysis, highlighting its potential to elevate research efficacy and outcomes. Read: https://lnkd.in/e7-dV6_f #AI #ChatGPT #Research

2

Check out the book "Machine Learning Systems - Principles and Practices of Engineering Artificially Intelligent Systems" by Vijay Janapa Reddi.

9