Hyeonho Song, Hard Negative Mixing for Contrastive Learning


Contrastive learning has become a key component of self-supervised learning approaches for computer vision. By learning to embed two augmented versions of the same image close to each other and to push the embeddings of different images apart, one can train highly transferable visual representations. As revealed by recent studies, heavy data augmentation and large

Sungwon Han, Improved protein structure prediction using potentials from deep learning


Protein structure prediction can be used to determine the three-dimensional shape of a protein from its amino acid sequence1. This problem is of fundamental importance as the structure of a protein largely determines its function2; however, protein structures can be difficult to determine experimentally. Considerable progress has recently been made by leveraging genetic information. It

Petter Holme, Temporal networks connecting the structure and function of socioeconomic systems


Abstract: Network science has developed into an essential interdisciplinary framework for discovering patterns in data, finding important nodes and edges, determining how to improve the resilience of systems, etc. Sometimes one knows not only what pairs of nodes are in contact but also when the contacts happen. To include such information about the timing of

Lexing Xie, The Anatomy of Online Video Popularity


Abstract: TBD Bio: Lexing Xie is Professor of Computer Science at the Australian National University, she leads the ANU Computational Media lab (http://cm.cecs.anu.edu.au). Her research interests are in machine learning, multimedia, social media. Of particular recent interest are stochastic time series models, neural networks for sequences and graphs. Her research is supported by the US

Axel Timmermann, Application of deep convolutional neural networks for detecting extreme weather in climate datasets


Abstract: Detecting extreme events in large datasets is a major challenge in climate science research. Current algorithms for extreme event detection are build upon human expertise in defining events based on subjective thresholds of relevant physical variables. Often, multiple competing methods produce vastly different results on the same dataset. Accurate characterization of extreme events in

Sujung Kim, Understanding the effects of tablet adoption and mobile news delivery on local news readership, subscriber satisfaction and retention


Abstract: While news organizations must create original news stories, there are other ways that they can provide value for their subscribers. One way is to use hardware devices and software interfaces to create a better reading experience. By offering a superior reading experience, subscribers will read more often, which will increase the perceived value of

Yong-Yeol Ahn, An implicit statistical bias of word2vec model and why it may be a good thing


Abstract: Neural language models have revolutionized how we model text data as well as a broad range of machine learning methods, even beyond methods for natural language processing. One of the first, simplest, and most widely used methods is the skip-gram negative sampling model, or simply word2vec, which allows us to obtain vector representations of

Joint AI+X & IBS Workshop on Data Science

IBS, Daejeon

This workshop, organized together with the AI+X forum at KAIST School of Computing, will present recent research on data science applied to large-scale datasets on healthcare, climate change, cultural analytics, econometrics, and computational biology. KAIST AI + X Forum on Meet the Data Scientists https://kaist.zoom.us/j/92330868763 10:00-12:00 (moderator: Sue Moon & Mia Cha) * Jisun An