Peter K F Cheung SBS’ Post

Foreign Asset To Efficiency Productivity Structure Foundation and Techniques for the migration group of technology and architecture and technology for the future of this process

We need to use tools to apply theoretical frameworks into practice. However, a good outcome is not assured by using tools. Which elements should we take into consideration to adapt our tools so they support our work to promote systems transformation? In Tricolor Coalition we were very happy to have co-designed a workshop Transformative Innovation Policy Consortium (TIPC) to learn together with other system thinkers of ways in which system thinking tools can be responsibly adapted to reflect different contexts and time challenges and opportunities. If you are interested to know the results of these workshop, click on the link below. #systemthinking #tipc #workshop #contextualization #theorytopractice

If the results in this paper hold/take off and the trade-offs are acceptable to people (which they seem reasonable to me) this could be game changing in networks for AI training and thus a big chunk of the total cost. Great paper Ying Zhang!

While studying the original LoRA paper, I was reminded of the present problem in adapting LLMs to downstream tasks, as stated by the authors: "As we pre-train larger models, full fine-tuning, which retrains all model parameters, becomes less feasible" and how LoRA is an alternative that should be expanded and continued to be worked on, as we look for even better solutions. Here are some of my notes if anyone wants to take a brief look at some things I highlighted in the Paper, that I think might convey many of the central ideas.

The transformational line of progress curved to form a circle: a middle way for technological advancements in journal production https://lnkd.in/eqNkXTf9 #CSEScienceEditor #scienceeditors #sciencepublishing

IWAIT 2025 strongly encourages the organization of Special Sessions to supplement regular programs, aiming to bring together leading researchers and practitioners from around the world to present their latest research in emerging topics of image technology. Special Session proposals should cover topics that relate to the regular technical program of IWAIT 2025, but are not limited to them. All submitted papers will go through the same review process, and decisions of acceptance will be based on the special session committee and the review results. The special session organizers and all invited speakers will be required to register for the conference. Submission of Special Session Proposal Proposals should include short summary with the following information: ‧ Name of the organizer(s) of the special session ‧ A title of the special session ‧ Session abstract (statement of the significance of the topic) ‧ Tentative titles and abstracts of at least 4 invited papers ‧ List of names who have agreed to contribute a paper in the session ‧ Short biography of special session organizer(s) that includes complete affiliation, address and email Pleae submit proposal by email Porf. Yi-Zeng Hsieh yzhsieh@mail.ntust.edu.tw

How to use the power of semantic technology to enable cross-disciplinary operations? Maybe worth to look into this event: https://lnkd.in/e6d9MDbz

The Transformative Innovation Policy Consortium (TIPC) has been applying systems thinking to develop tools that can allow us to create transformational change across different sustainability projects. In this article, we share our experience employing the Transformative Theory of Change tool.

Shedding light on the requirements to extend AIs and class architectures….LLM, Expert Systems and knowledge management/declaration. Great article.

Check out the latest publication from the IDEASLab, led by Zeynep Duygu Tekler! In this paper, we tackle the data-intensive challenge of creating personal thermal comfort models for occupant-centric controls by proposing a data-efficient framework that leverages both active learning and transfer learning. In the paper, we demonstrate that our approach outperforms baselines that include other transfer learning models using less than 10% of the training data. Access the full paper with the 50-day free access share link below. Code and data are also available via a research compendium hosted on GitHub (link in paper). Co-authors: Zeynep Duygu Tekler, Yue Lei, Adrian Chong Share link: https://lnkd.in/gizJFAVN