Convergence of machine learning with microfluidics and metamaterials to build smart materials

Recent advances in machine learning have revolutionized numerous research domains by extracting the hidden features and properties of complex systems, which are not otherwise possible using conventional ways. One such development can be seen in designing smart materials, which intersects the ability of microfluidics and metamaterials with machine learning to achieve unprecedented abilities. Microfluidics involves generating and manipulating fluids in the form of liquid streams or droplets from microliter to femtoliter regimes. However, analysis of such fluid flows is always tiresome and challenging due to the complexity involved in the integration and detection of various chemical or biological processes. On the other hand, acoustic metamaterials manipulate acoustic waves to achieve unparalleled properties, which is not possible using natural materials. Nonetheless, the design of such metamaterials relies on the expertise of specialists or on analytical models that require an enormous number of expensive function evaluations, making this method extremely complex and time-consuming. These complexities and exorbitant function evaluations of both fluidic and metamaterial systems embark on the need for the support of computational tools that can identify, process, and quantify the large amounts of intricacy, thus machine learning techniques. This review discusses the shortcomings of microfluidics and acoustic metamaterials, which are overcome by neoteric machine learning approaches for building smart materials. The following review ends by providing the importance and future perspective of integrating machine learning and optimization approaches with microfluidic-based acoustic metamaterials to build smart and efficient intelligent next-generation materials.