有興趣參加者，敬請於11/25(一)前完成線上報名(https://tinyurl.com/2bqvcfev)。
活動主題(Subject)：2024諾貝爾物理獎之前世今生
日 期 (Date): November 28, 2024 (Thu.)
地 點 (Place): 物理所1樓演講廳 1F, Auditorium, Institute of Physics
時 間 (Time): 14:00-15:10
演講者 (Speaker): 羅中泉教授 (Professor Chung-Chuan Lo)
所屬機構 (Institute): 國立清華大學系統神經科學研究所 (Institute of Systems Neuroscience, National Tsing Hua University)
題 目 (Title): 從伽伐尼到霍普菲爾德：兩世紀的計算神經科學旅程如何奠定了現代人工智慧 (From Galvani to Hopfield: How A Two-Century Journey in Neural Computation Leading to Modern AI)
摘 要(Abstract):
在過去的兩個世紀中，我們在神經計算的知識上的演進為現代人工智慧奠定了基礎。本演講將從路易吉·伽伐尼（Luigi Galvani）於1780年發現神經傳遞電訊號開始，追溯關鍵的里程碑，包括霍奇金-赫胥黎模型（Hodgkin-Huxley model）——艾倫·霍奇金（Alan Hodgkin）和安德魯·赫胥黎（Andrew Huxley）因此獲得了1963年的諾貝爾生理學或醫學獎——該模型數學地描述了神經元如何產生和傳播動作電位。 接下來，我們將探討從單一神經元到神經網路的理論科學演進，除了介紹幾個重要的理論與模型，重點將放在介紹由約翰·霍普菲爾德（John J. Hopfield）提出的霍普菲爾德網路。霍普菲爾德與傑佛瑞·辛頓（Geoffrey Hinton）共同獲得了2024年諾貝爾物理學獎。本次演講將帶大家瀏覽這段非凡的科學旅程，介紹那些導致現代人工智慧的關鍵發展，並向今年諾貝爾獎所表彰的先驅者致敬。 接下來，我們將探討從單一神經元到神經網路的理論科學演進，除了介紹幾個重要的理論與模型，重點將放在介紹由約翰·霍普菲爾德（John J. Hopfield）提出的霍普菲爾德網路。霍普菲爾德與傑佛瑞·辛頓（Geoffrey Hinton）共同獲得了2024年諾貝爾物理學獎。本次演講將帶大家瀏覽這段非凡的科學旅程，介紹那些導致現代人工智慧的關鍵發展，並向今年諾貝爾獎所表彰的先驅者致敬。
Over the past two centuries, our understanding of neural computation has evolved dramatically, laying the foundation for modern artificial intelligence. Beginning with Luigi Galvani’s 1780 discovery that nerves transmit electrical signals, this lecture traces key milestones, including the Hodgkin-Huxley model—which earned Alan Hodgkin and Andrew Huxley the 1963 Nobel Prize in Physiology or Medicine—for mathematically describing how neurons generate and propagate action potentials. We will then explore the evolution from individual neurons to neural networks, introducing several key theories and models and highlighting the Hopfield network, a groundbreaking contribution of John J. Hopfield, who shared the 2024 Nobel Prize in Physics with Geoffrey Hinton. This lecture celebrates this remarkable journey, underscoring the pivotal developments that have led to today’s artificial intelligence and honoring the pioneers recognized by this year’s Nobel Prize.
連 結 (Link): http://www.phys.sinica.edu.tw/lecture_detail.php?id=2965&eng=T
演講語言 (Language): in English
時 間 (Time): 15:30-16:40
演講者 (Speaker): 王道維教授 (Prof. Daw-Wei Wang)
所屬機構 (Institute): 國立清華大學物理系/人文社會AI應用與發展研究中心 (Department of Physics and Research Center for Applications and Development of AI in Humanities and Social Sciences, National Tsing Hua University)
題 目 (Title): 當人造虛擬進入物理真實 (When Artificial Virtuality Enters Physical Reality)
摘 要(Abstract):
2024年諾貝爾物理獎頒給AI教父Geoffrey Hinton，引發了許多爭議。Hinton雖非物理學家，但他所推動的類神經網路技術在各領域大放異彩，使原本屬於資工演算法的機器學習，意外被視為物理研究的成果而得獎。此次我將先簡要回顧Hinton的求學和研究歷程，以基礎的物理概念來說明類神經網路的本質，並解釋後來如何演化成為當今人工智慧領域最成功的演算法—深度學習。在簡介幾個應用於物理研究的例子後，我也將說明AI的應用與其潛在的問題如何對人文社會領域帶來重大的影響。這是為何Hinton於2023年選擇離開Google、對自己發明深度學習表達後悔，反而積極呼籲暫緩當前的AI發展，免得危及人類自身。面對AI這個人類歷史上最具跨領域且顛覆性的技術，這次的諾貝爾物理獎在科學史上顯然具有典範性的獨特意義。
The 2024 Nobel Prize in Physics was awarded to Geoffrey Hinton, a pioneer of AI, sparking significant controversy. Although Hinton is not a physicist, the neural network technology he advanced has achieved remarkable success across various fields, leading machine learning—originally a branch of computer science—to be unexpectedly recognized as a contribution to physics research. In this talk, I will begin with a brief overview of Hinton’s educational and research journey, explaining the essence of neural networks using fundamental physics concepts. I will then explore how these networks evolved into deep learning, now the most successful algorithm in artificial intelligence. After introducing several examples of AI applications in physics research, I will discuss the profound impact—and potential risks—of AI on the humanities and social sciences. This context helps explain why Hinton left Google in 2023, expressed regret over his role in creating deep learning, and began advocating for a slowdown in AI development to safeguard humanity. As one of the most interdisciplinary and disruptive technologies in human history, AI’s recognition with a Nobel Prize in Physics clearly represents a paradigmatic milestone in the history of science.
連 結 (Link):
http://www.phys.sinica.edu.tw/lecture_detail.php?id=2964&eng=T
https://phys.site.nthu.edu.tw/p/406-1335-58679,r3581.php?Lang=zh-tw
演講語言 (Language): in Mandarin
接待人 (Host): 張元翰(Chang, Yuan-Hann) 博士
聯絡人 (Contact): 洪敏玲 #6750
—————————————————————-
Colloquium-Seminar
Institute of Physics, Academia Sinica
128, Sec. 2, Academia Road, Nankang,Taipei 115-29,Taiwan, R.O.C
Tel: +866-2-2789-8365
E-mail: colloquium-seminar@phys.sinica.edu.tw