To tackle the high energy and latency costs of compressed sensing workloads in edge computing, researchers at Tsinghua University developed a memristor-based compressed sensing accelerator (memCS). By utilizing a computing-in-memory (CIM) architecture and hardware-software co-optimization framework to mitigate accuracy loss from hardware non-idealities, the memCS achieved a near-software computing accuracy (31.11 dB peak signal-to-noise ratio) while delivering an 11.22x speedup and 30.46x energy savings compared to GPUs, paving the way for efficient edge computing.

This study presents a novel approach to natural antimicrobial skincare by formulating herbal creams using fresh (FBP) and dried banana peel (DBP) extracts. Recognizing the rich bioactive and lignocellulosic compounds in banana peels, the researchers explored their potential as eco-friendly alternatives to synthetic antimicrobial agents in cosmetic products. The FBP-based cream demonstrated strong antibacterial activity, particularly against Staphylococcus aureus and Pseudomonas aeruginosa, with inhibition zones reaching 40.73 mm. Conversely, the DBP-based cream exhibited superior antifungal efficacy against Candida albicans and Aspergillus niger, with zones up to 49.63 mm. Physical and cosmetic properties were carefully assessed, revealing that FBP creams had better spreadability, uniformity, and texture compared to DBP formulations. The study indicates that combining both fresh and dried extracts could provide broad-spectrum antimicrobial protection. Overall, the research demonstrates that banana peel extracts are not only sustainable and low-cost but also highly effective in topical applications, bridging natural waste valorization with innovative cosmetic product development.