Machine learning technologies enable such intelligent judgement as classification, prediction and anomaly detection based on computer's learning from past cases and observation data. They are being widely used in various kinds of problem-solving, coupled with increase and availability of big data. This article describes four technological challenges important for the problem-solving: (1) to improve explainability of learned results, (2) to develop a total framework to integrate machine learning with decision-making, (3) to improve speed and efficiency of deep learning, and (4) to establish a new methodology for inductive development of machine learning-based systems.

The genetic information of a living organism is encoded on the genomic DNA. To fully understand the phenomenon of life, it is critical to elucidate how genes are expressed as required. In addition to the regulation by transcription factors that bind to specific DNA sequence, epigenetic histone medications and chromatin conformation in the nucleus are also crucial. Recently, "Nucleome" research has been highlighted, as the gene regulation can only be addressed in the context of cell nucleus in space and time. From the aspect of "Nucleome", integrated analyses of microscopy, genome biology, and informatics and mathematical sciences would be facilitated. The 4D Nucleome Program has already been funded in the US, and the 4D Nucleome Initiative in Europe has launched. There is a movement of launching "The International Nucleome Consortium". "Nucleome" research should be facilitated also in Japan.

Biometrics technology has been utilized in various scenes as individual authentication method. NEC's face recognition technology gets four consecutive 1st rank in US National Institute of Standards and Technology tests. There are two authentication methods in face authentication, one is cooperative and another is non-cooperative. The latter method is particularly noteworthy now. It is used in the case that the target is not conscious of being authenticated or we cannot get agree of target. Video face recognition technology is used in this non-cooperative authentication. In this authentication, it is important to match faces from low resolution images. And NEC acquired the highest evaluation in the test of NIST. The problem of face recognition is to present the list of faces, but new technologies to solve the problem have been developed. So, face recognition will be utilized more commonly.

Quantification of symptom severity, without question, is important in various medical fields. However, due to the lack of biomarkers that closely reflect illness severity in psychiatry, the field suffers greatly in diagnosing, assessing treatment response, and developing new drugs. Utilization of information technology and/or machine learning may be one of the promising approaches to resolve this problem. In psychiatry, centralized rating via live two-way video is often used to eliminate raters’ biases in clinical trials. However, rating scales themselves have their limitations in validity and reliability. Thanks to the progress in information communication technology (ICT), there are multiple projects across the world utilizing such technology to assist diagnosing and/or quantifying the severity. Such attempts include conversation analysis using smartphone, activity monitoring by wearable devices. PROMPT (Project for Objective Measures Using Computational Psychiatry Technology), which launched in November of 2015, is a government-funded study through AMED (Japan Agency for Medical Research and Development).While Keio university is leading the project, seven companies have participated in this project and are contributing their technologies. The goal of the project is to quantify the severity of psychiatric symptoms.Another project named UNDERPIN (Understanding Psychiatric Illness through Natural Language Processing) is a collaborative study between Keio university and Shizuoka university to extract features of psychiatric diseases from written and spoken language, utilizing natural language processing. The goal of the project is to identify early sign of disease and to facilitate prevention and early intervention.