Embedded Files
From "Reminder of the First Paper on Transfer Learning in Neural Networks, 1976" by S. Bozinovski
Informatica 44.3 (2020)
About
About
In 1976, S. Bozinovski and A. Fulgosi published one of the earliest papers on transfer learning in neural networks, "The influence of pattern similarity and transfer learning upon training of a base perceptron b2" (in Croatian originally) in the Proceeding of Informatica. They provided a mathematical model for this particular transfer learning problem and explained both "positive" and "negative" transfer from a geometric point of view. Since then, transfer learning has been widely adopted in cognitive science and computer science. The recent rise of Large Language Models, to a great extent, relies on various transfer learning techniques. Besides, transfer learning has also been gaining its popularity in the field of finance. In the meantime, researchers never stop the study on "positive" and "negative" transfer and have proposed various theoretical insights and quantitative measurements. In this workshop, we will present these current studies on transfer learning and introduce some of its applications in finance.
In 1976, S. Bozinovski and A. Fulgosi published one of the earliest papers on transfer learning in neural networks, "The influence of pattern similarity and transfer learning upon training of a base perceptron b2" (in Croatian originally) in the Proceeding of Informatica. They provided a mathematical model for this particular transfer learning problem and explained both "positive" and "negative" transfer from a geometric point of view. Since then, transfer learning has been widely adopted in cognitive science and computer science. The recent rise of Large Language Models, to a great extent, relies on various transfer learning techniques. Besides, transfer learning has also been gaining its popularity in the field of finance. In the meantime, researchers never stop the study on "positive" and "negative" transfer and have proposed various theoretical insights and quantitative measurements. In this workshop, we will present these current studies on transfer learning and introduce some of its applications in finance.
Speakers
Speakers
Haoyang Cao - École Polytechnique
Álvaro Cartea - University of Oxford
Shao-Lun Huang - Tsinghua-Berkeley Shenzhen Institute
Xinyu Li - University of California, Berkeley
Justing Sirignano - University of Oxford
Kaizheng Wang - Columbia University
Organizers: Haoyang Cao, Haotian Gu and Xin Guo
Page updated
Google Sites
Report abuse