Autonomous materials research labs require the ability to combine and learn from diverse data streams. This is especially true for learning material synthesis-process-structure-property relationships, key to accelerating materials optimization and discovery as well as accelerating mechanistic understanding. We present the Synthesis-process-structure-property relAtionship coreGionalized lEarner (SAGE) algorithm. A fully Bayesian algorithm that uses multimodal coregionalization to merge knowledge across data sources to learn synthesis-process-structure-property relationships. SAGE outputs a probabilistic posterior for the relationships including the most likely relationships given the data.

Graph Neural Networks (GNNs) as deep learning models working on graph-structure data have achieved advanced performance in many works. However, it has been proved repeatedly that, not all edges in a graph are necessary for the training of machine learning models. In other words, some of the connections between nodes may bring redundant or even misleading information to downstream tasks. In this paper, we try to provide a method to drop edges in order to purify the graph data from a new perspective. Specifically, it is a framework to purify graphs with the least loss of information, under which the core problems are how to better evaluate the edges and how to delete the relatively redundant edges with the least loss of information. To address the above two problems, we propose several measurements for the evaluation and different judges and filters for the edge deletion. We also introduce a residual-iteration strategy and a surrogate model for measurements requiring unknown information. The experimental results show that our proposed measurements for KL divergence with constraints to maintain the connectivity of the graph and delete edges in an iterative way can find out the most edges while keeping the performance of GNNs. What's more, further experiments show that this method also achieves the best defense performance against adversarial attacks.

Many FinTech companies incorporate various methods to distinguish fraud from ordinary transactions. But it is even better to prevent fraud even before it happens. Each founder of a FinTech startup has to remember that it is impossible to prevent fraud once and for all. Your task is to prevent it from scaling. And this is a moment where technology kicks in. What are other AI-powered methods to prevent fraud in FinTech startups and companies?

Artificial Intelligence is a hot topic today, and while there are some groups who claim that another winter may be coming, a larger population (including myself) strongly feel that this time, summer is here and it's going to be one big party. In fact, with advances in both hardware and software, there may not be winter in sight for a long time. Below are the top 10 trends I am excited about in 2020. Towards the end of 2019, Google's announcement of quantum computing power, which outperformed a standard supercomputer by a factor of over a billion, caused waves in the media. While there may not be any direct use for it in real-world applications today, there is extensive focus on quantum computing in research labs at companies such as Google and IBM.

Personalizing the News Feed: A Large-Scale Recommendation Problem Personalization is a key component in ensuring user satisfaction, and at Yahoo, personalization is at the heart of several user-facing products. This talk will focus on how Yahoo built one of the largest news recommendation engines in the world: the Yahoo stream, which personalizes the news feed for several hundreds of millions of users on millions of content items. Beyond the scale, the success of the news feed also depends on whether it is able to engage the user long term. In this session, Yahoo's director of research will present the challenges and issues in designing an engaging stream, and attendees will also learn how to cope with sparsity of explicit feedback, how user behavior changes with context of the device, how to build machine learned models for each user, and the metric that allows Yahoo to optimize for long term user-engagement.