DLC 2018 - Deep Learning and Clustering (DLC)

Roughly, clustering is the process of organizing similar objects into meaningful clusters. This approach is essential in many fields, including data science, machine learning, information retrieval, bio-informatics and computer vision, to deal with massive data. Despite their success, most existing clustering methods are severely challenged by the data generated by modern applications, which are typically high dimensional, noisy, heterogeneous and sparse, etc. This has driven many researchers to investigate new clustering models to overcome these difficulties. One promising category of such models relies on data embedding. The idea is to embed the original data into a low dimensional latent space and then perform clustering on this new space. The goal of the embedding here is to learn new representations of the objects of interest, e.g., images, that encode only the most relevant information characterizing the original data, which would for example reduce noise and sparsity. Since the embedding process is not guaranteed to infer representations that are suitable for the clustering task, it is important to perform both tasks jointly, as recommended by several authors, so as to let clustering govern feature extraction and vice-versa. Within this framework, classical dimensionality reduction approaches, e.g., Principal Component Analysis (PCA), have been widely considered for the embedding task. However, the linear nature of such techniques makes it challenging to infer faithful representations of real-world data, which typically lie on highly non-linear manifolds. This motivates the investigation of deep learning models (e.g., auto-encoders, convolutional neural networks, etc.), which have proven so far successful in extracting highly non-linear features from complex data, such as text, images, graphs, etc. While promising, composing deep learning with clustering simultaneously (referred to as deep clustering) has just started.
Hence, one main goal of the workshop is to bring together the leading researchers who work on state-of-the-art deep unsupervised feature extraction and clustering models, and also the practitioners who seek for novel applications. In summary, this workshop is an opportunity to:
Present the recent advances in deep learning based clustering algorithms.
Outline potential applications that could inspire new deep clustering approaches.
Explore benchmark data to better evaluate and study deep clustering models.