Machine learning security has recently become a prominent topic in the natural language processing (NLP) area. The existing black-box adversarial attack suffers prohibitively from the high model ...querying complexity, resulting in easily being captured by anti-attack monitors. Meanwhile, how to eliminate redundant model queries is rarely explored. In this paper, we propose a query-efficient approach BufferSearch to effectively attack general intelligent NLP systems with the minimal number of querying requests. In general, BufferSearch makes use of historical information and conducts statistical test to avoid incurring model queries frequently. Numerically, we demonstrate the effectiveness of BufferSearch on various benchmark text-classification experiments by achieving the competitive attacking performance but with a significant reduction of query quantity. Furthermore, BufferSearch performs multiple times better than competitors within restricted query budget. Our work establishes a strong benchmark for the future study of query-efficiency in NLP adversarial attacks.
As one of the most powerful topic models, Latent Dirichlet Allocation (LDA) has been used in a vast range of tasks, including document understanding, information retrieval and peer-reviewer ...assignment. Despite its tremendous popularity, the security of LDA has rarely been studied. This poses severe risks to security-critical tasks such as sentiment analysis and peer-reviewer assignment that are based on LDA. In this paper, we are interested in knowing whether LDA models are vulnerable to adversarial perturbations of benign document examples during inference time. We formalize the evasion attack to LDA models as an optimization problem and prove it to be NP-hard. We then propose a novel and efficient algorithm, EvaLDA to solve it. We show the effectiveness of EvaLDA via extensive empirical evaluations. For instance, in the NIPS dataset, EvaLDA can averagely promote the rank of a target topic from 10 to around 7 by only replacing 1% of the words with similar words in a victim document. Our work provides significant insights into the power and limitations of evasion attacks to LDA models.
Polymeric micelles (PM) system, as an efficient drug carrier, has received growing scientific attention in recent years owing to its solubilization, selective targeting, P-glycoprotein inhibition and ...altered drug internalization route and subcellular localization properties. Seven PM formulations of anti-tumor drugs being evaluated in clinical trials are reviewed in this paper, in terms of formulation study, in vitro cytotoxicity, in vivo pharmacokinetics, anti-tumor efficacy and safety as well as clinical trials, to shed new light on the discovery of novel PM formulations. In these seven PM formulations, PM system was employed to overcome the issues of low water solubility, high toxicity and (or) multidrug resistance accompanied with the conventional formulation, which greatly hampered their clinical application. Those promising preclinical and clinical results combined with rapid advancement and intense multidisciplinary collaboration enable the extension of the PM system to traditional Chinese medicine, imaging agents, gene and combination agent deliveries as well as some other administration routes, which facilitate the clinical translation of the PM drug delivery system.
Display omitted
Calorie restriction (CR) and fasting are common approaches to weight reduction, but the maintenance is difficult after resuming food consumption. Meanwhile, the gut microbiome associated with energy ...harvest alters dramatically in response to nutrient deprivation. Here, we reported that CR and high-fat diet (HFD) both remodeled the gut microbiota with similar microbial composition, Parabacteroides distasonis was most significantly decreased after CR or HFD. CR altered microbiota and reprogramed metabolism, resulting in a distinct serum bile acid profile characterized by depleting the proportion of non-12α-hydroxylated bile acids, ursodeoxycholic acid and lithocholic acid. Downregulation of UCP1 expression in brown adipose tissue and decreased serum GLP-1 were observed in the weight-rebound mice. Moreover, treatment with Parabacteroides distasonis or non-12α-hydroxylated bile acids ameliorated weight regain via increased thermogenesis. Our results highlighted the gut microbiota-bile acid crosstalk in rebound weight gain and Parabacteroides distasonis as a potential probiotic to prevent rapid post-CR weight gain.
Due to the inherent resistance of bacterial biofilms to antibiotics and their serious threat to global public health, novel therapeutic agents and strategies to tackle biofilms are urgently needed. ...To this end, we designed and synthesized a novel guanidinium‐functionalized pillar5arene (GP5) that exhibited high antibacterial potency against Gram‐negative E. coli (BH101) and Gram‐positive S. aureus (ATCC25904) strains. More importantly, GP5 effectively disrupted preformed E. coli biofilms by efficient penetration through biofilm barriers and subsequent destruction of biofilm‐enclosed bacteria. Furthermore, host–guest complexation between GP5 and cefazolin sodium, a conventional antibiotic that otherwise shows negligible activity against biofilms, exhibited much enhanced, synergistic disruption activity against E. coli biofilms, thus providing a novel supramolecular platform to effectively disrupt biofilms.
Guanidinium‐functionalized pillar5arene (GP5) exhibited antibacterial activity against both Gram‐negative E. coli and Gram‐positive S. aureus bacterial strains. More significantly, it showed strong biofilm‐disrupting activity against preformed E. coli biofilms. Host–guest complexation between GP5 and a conventional antibiotic, cefazolin sodium, provides a supramolecular strategy for synergistically enhanced disruption of bacterial biofilms (see picture).
Hyocholic acid (HCA) is a major bile acid (BA) species in the BA pool of pigs, a species known for its exceptional resistance to spontaneous development of diabetic phenotypes. HCA and its ...derivatives are also present in human blood and urine. We investigate whether human HCA profiles can predict the development of metabolic disorders. We find in the first cohort (n = 1107) that both obesity and diabetes are associated with lower serum concentrations of HCA species. A separate cohort study (n = 91) validates this finding and further reveals that individuals with pre-diabetes are associated with lower levels of HCA species in feces. Serum HCA levels increase in the patients after gastric bypass surgery (n = 38) and can predict the remission of diabetes two years after surgery. The results are replicated in two independent, prospective cohorts (n = 132 and n = 207), where serum HCA species are found to be strong predictors for metabolic disorders in 5 and 10 years, respectively. These findings underscore the association of HCA species with diabetes, and demonstrate the feasibility of using HCA profiles to assess the future risk of developing metabolic abnormalities.
In this paper, a stochastic model predictive control (MPC) method based on reinforcement learning is proposed for energy management of plug-in hybrid electric vehicles (PHEVs). Firstly, the power ...transfer of each component in a power-split PHEV is described in detail. Then an effective and convergent reinforcement learning controller is trained by the Q-learning algorithm according to the driving power distribution under multiple driving cycles. By constructing a multi-step Markov velocity prediction model, the reinforcement learning controller is embedded into the stochastic MPC controller to determine the optimal battery power in predicted time domain. Numerical simulation results verify that the proposed method achieves superior fuel economy that is close to that by stochastic dynamic programming method. In addition, the effective state of charge tracking in terms of different reference trajectories highlight that the proposed method is effective for online application requiring a fast calculation speed.
•Stochastic model predictive control is achieved based on reinforcement learning.•The Q-learning algorithm is employed to build the reinforcement learning controller.•A multi-step Markov velocity prediction model is embedded into the controller.•The proposed method achieves superior fuel economy with fast calculation speed.
Platinum-based heterogeneous catalysts are critical to many important commercial chemical processes, but their efficiency is extremely low on a per metal atom basis, because only the surface ...active-site atoms are used. Catalysts with single-atom dispersions are thus highly desirable to maximize atom efficiency, but making them is challenging. Here we report the synthesis of a single-atom catalyst that consists of only isolated single Pt atoms anchored to the surfaces of iron oxide nanocrystallites. This single-atom catalyst has extremely high atom efficiency and shows excellent stability and high activity for both CO oxidation and preferential oxidation of CO in H2. Density functional theory calculations show that the high catalytic activity correlates with the partially vacant 5d orbitals of the positively charged, high-valent Pt atoms, which help to reduce both the CO adsorption energy and the activation barriers for CO oxidation.