VTechWorks staff will be away for the Thanksgiving holiday beginning at noon on Wednesday, November 27, through Friday, November 29. We will resume normal operations on Monday, December 2. Thank you for your patience.

Browsing by Author "Wang, Tianyi"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Effects of Individual Essential Amino Acids on Growth Rates of Young Rats Fed a Low-Protein Diet
    Liu, Wei; Wang, Tianyi; Zhao, Kai; Hanigan, Mark D.; Lin, Xueyan; Hu, Zhiyong; Hou, Qiuling; Wang, Yun; Wang, Zhonghua (MDPI, 2024-03-20)
    To investigate the effects of individual essential amino acids (EAA) on growth and the underlying mechanisms, EAA individually supplemented a low-protein (LP) diet fed to young rats in the present study. Treatments were an LP diet that contained 6% crude protein (CP), a high-protein (HP) diet that contained 18% CP, and 10 LP diets supplemented with individual EAA to achieve an EAA supply equal to that of the HP diet. The CP concentration of the LP diet was ascertained from the results of the first experiment, which examined the effects of dietary CP concentrations on growth rates, with CP ranging from 2% to 26%. Weight gain was increased with the supplementation of His, Ile, Lys, Thr, or Trp as compared to the LP diet (p < 0.05). Feed intake was greater for the His-, Lys-, and Thr-supplemented treatments as compared to the LP group (p < 0.05). Protein utilization efficiency was lower for the HP group than other groups (p < 0.01). The supplementation of Leu, Lys, and Val led to reduced protein utilization efficiency (p < 0.05), but the supplementation of Thr and Trp led to greater efficiency than the LP group (p < 0.05). Compared to the LP group, plasma urea concentrations were elevated with individual EAA supplementation, with the exception of the Thr addition. The added EAA resulted in increased concentrations of the corresponding EAA in plasma, except for Arg and Phe supplementation. The supplementation of Arg, His, Leu, Lys, and Met individually stimulated mTORC1 pathway activity (p < 0.05), and all EAA resulted in the decreased expression of ATF4 (p < 0.05). In summary, the supplementation of His, Ile, Lys, Thr, or Trp to an LP diet improved the growth performance of young rats. Responses to His and Lys additions were related to the activated mTORC1 pathway and feed intake increases. The improved growth performance resulting from the addition of a single EAA is not solely attributed to the increased plasma availability of EAA. Rather, it may be the consequence of a confluence of factors encompassing signaling pathways, the availability of amino acids, and other associated elements. The additivity of these factors results in independent responses to several EAA with no order of limitation, as is universally encoded in growth models for all production animal species.
  • Thumbnail Image
    Man vs. Machine: Practical Adversarial Detection of Malicious Crowdsourcing Workers
    Gang, Wang; Wang, Tianyi; Zheng, Haitao; Zhao, Ben Y. (USENIX, 2014-08)
    Recent work in security and systems has embraced the use of machine learning (ML) techniques for identifying misbehavior, e.g. email spam and fake (Sybil) users in social networks. However, ML models are typically derived from fixed datasets, and must be periodically retrained. In adversarial environments, attackers can adapt by modifying their behavior or even sabotaging ML models by polluting training data. In this paper¹, we perform an empirical study of adversarial attacks against machine learning models in the context of detecting malicious crowdsourcing systems, where sites connect paying users with workers willing to carry out malicious campaigns. By using human workers, these systems can easily circumvent deployed security mechanisms, e.g. CAPTCHAs. We collect a dataset of malicious workers actively performing tasks on Weibo, China’s Twitter, and use it to develop MLbased detectors. We show that traditional ML techniques are accurate (95%–99%) in detection but can be highly vulnerable to adversarial attacks, including simple evasion attacks (workers modify their behavior) and powerful poisoning attacks (where administrators tamper with the training set). We quantify the robustness of ML classifiers by evaluating them in a range of practical adversarial models using ground truth data. Our analysis provides a detailed look at practical adversarial attacks on ML models, and helps defenders make informed decisions in the design and configuration of ML detectors.