We have lately witnessed one-after-another ground-breaking developments in the AI world. Data science is the essence of this world and perhaps the fastest expanding domain. At the same time, data science is born from the interaction of multiple research areas including statistics and machine learning. The fundamental statistical principles formed a pillar of data science in the beginning and remain pivotal for its healthy growth. The powerful algorithms developed in computer science keep data science practically relevant and central in Big Data analytics. Promoting the interactions between these two areas will undoubtedly further advance the development of data science. Latent structure models are powerful products of statistical principles and offer a simple yet effective platform to capture the complexity and heterogeneity in big data. Their usage goes beyond any single discipline, and naturally integrates the advantages of various areas such as statistics, economics and machine learning. This workshop brings together researchers with expertise in latent structure learning and provides a platform to share their achievements, exchange research ideas, and build new collaborations. We wish to promote the use of latent model techniques and simulate further development of highly interpretable, reproducible, and powerful AI methods. The workshop covers a broad range of topics in the latent structure model: theory, algorithms, and applications. We will also set eyes on future research trends and interdisciplinary directions.