Search

In e-commerce, content quality of the product catalog plays a key role in delivering a satisfactory experience to the customers. In particular, visual content such as product images influences customers' engagement and purchase decisions. With the rapid growth of e-commerce and the advent of artificial intelligence, traditional content management systems are giving way to automated scalable systems. In this paper, we present a machine learning driven visual content management system for extremely large e-commerce catalogs. For a given product, the system aggregates images from various suppliers, understands and analyzes them to produce a superior image set with optimal image count and quality, and arranges them in an order tailored to the demands of the customers. The system makes use of an array of technologies, ranging from deep learning to traditional computer vision, at different stages of analysis. In this paper, we outline how the system works and discuss the unique challenges related to applying machine learning techniques to real-world data from e-commerce domain. We emphasize how we tune state-of-the-art image classification techniques to develop solutions custom made for a massive, diverse, and constantly evolving product catalog. We also provide the details of how we measure the system's impact on various customer engagement metrics., Comment: Accepted in IEEE Big Data Conference 2018 (Industry & Government Track)

The aim of this thesis is twofold: on the one hand to demonstrate a new method of building magnetic nanostructures at surfaces and on the other hand to present the construction of a new 1K UHV Low Temperature STM aimed at coupling the spatial resolution of the STM microscope with the capability to detect magnetism. The focus of the first chapter is the construction of the entire system incorporating the STM microscope. Design and the functionalities of the instrument are emphasized. The system is an Eigler type UHV LT STM. This class of low T STM relies on a pendulum mechanism which mechanically decouples the STM from the entire UHV system. As a result the mechanical stability of the microscope is particularly improved. Besides, this design will allow eventually to develop a temperature control allowing the instrument to work between 1 and 80 K. Unlike the first generation of Eigler type STM, the temperature can be lowered below 4.2 K by the use of a Joule-Thompson expansion closed cycle. He3 is let to expand into the cavity allocating the pendulum. Upon circulation of the gas around the pendulum the microscope is cooled down to a temperature as low as 0.5 K. The STM microscope itself is adapted from a design originally proposed by Pan and then developed by Pietzch. This design is compact, stable, provides a good thermal stability and allows for in situ 180° sample rotation. Different proposals for the detection of magnetism with the STM are described at the end of the chapter. The second chapter is inherently focused on the construction of magnetic nanostructures at surfaces under UHV conditions. A VT-STM is used to characterize organo-metallic assemblies formed on Cu(100). A magnetic atom, Fe, and an organic molecule, TMA, are deposited on the surface. A chemical reaction between the two adsorbates takes place. The TMA molecules offers three carboxyl moieties for lateral linkage to the Fe atoms. In this respect the coordinative interaction between the carboxyl moi