When learning in a domain with few training data, a method to improve learning efficiency by using training data in other available domains is called transfer learning. In this presentation, we focus on transfer learning in the case where data cannot be shared between domains, and explain a learning algorithm that provides theoretical guarantees for negative transfer. As a concrete application, we describe a reading support system using the proposed method for the problem of detecting cerebral aneurysms between different hospitals.

　In recent years, it has become clear that RNA molecules play a central role in the expression of higher biological phenomena such as development, differentiation, and cancer. I will introduce the past and present research on RNA.

　Currently, the world has a population of more than 7 billion people. In the future, rapid population growth will occur mainly in developing countries such as sub-Saharan Africa, and the world population is expected to reach more than 9 billion by 2050. The world's population is expected to reach more than 9 billion by 2050, which means that food must be produced so that "all people, at all times, have physical and economic access to sufficient, safe and nutritious food to meet their dietary needs and preferences for an active and healthy life. On the other hand, it has been noted that climate change is already affecting food security, and the threat is expected to intensify in the future. Needless to say, it is the rural areas of developing countries where food security is most vulnerable. In this seminar, we will discuss the food supply and demand and poverty alleviation in developing countries by reviewing the survey research conducted in rural areas of developing countries, taking into account the recent criticisms and research trends on food and agricultural policies that have continued since the "Green Revolution".