This study investigated the physicochemical properties, functionalities, and antioxidant capacities of protein extracts from wild sea cucumber Australostichopus mollis collected from four distinct locations in New Zealand. Protein was extracted from sea cucumber body walls using trypsin enzymatic extraction, followed by cold acetone precipitation. The amino acid analysis revealed high glycine (189.08 mg/g), glutamic acid (119.45 mg/g), and aspartic acid (91.91 mg/g) concentrations in all samples. The essential amino acid indexes of the protein extracts (62.96, average) were higher than the WHO/FAO standard references, indicating the excellent protein quality of A. mollis. Furthermore, protein extracts from A. mollis demonstrated superior emulsifying activity (202.3–349.5 m²/g average) compared to commercial soy and whey protein isolates under all tested pH conditions, and enhanced foaming capacity (109.9–126.4%) and stability (52.7–72%) in neutral and acidic conditions. The extracts also exhibited good solubility, exceeding 70% across pH 3–11. Antioxidant capacities (ABTS and DPPH free radical scavenging activity and ferric reducing antioxidant power) were identified in A. mollis protein extracts for the first time, with clear variations observed among different locations. These findings elucidate the advantageous functional properties of protein extracts from wild New Zealand A. mollis and highlight their potential application as high-quality antioxidant food ingredients.

In recent years, deep neural networks (DNNs) have emerged as the most successful technology for many difficult problems in image, video, voice and text processing. DNNs are resource hungry and require very large amounts of computation and memory, which is a particular challenge on IoT, mobile and embedded systems. In this chapter, we outline some major performance challenges of DNNs such as computation, parallelism, data locality and memory requirements. We describe research on these problems, such as the use of existing high-performance linear algebra libraries, hardware acceleration, reduced-precision storage and arithmetic and sparse data representations. Finally, we discuss recent trends in adapting compiler and domain-specific program generation techniques to create high-performance parallel DNN programs.