Traffic flow prediction is crucial for urban traffic management and public safety. Its key challenges lie in how to adaptively integrate the various factors that affect the flow changes. In this paper, we propose a unified neural network module to address this problem, called Attentive Crowd Flow Machine (ACFM), which is able to infer the evolution of the crowd flow by learning dynamic representations of temporally-varying data with an attention mechanism. Specifically, the ACFM is composed of two progressive ConvL-STM units connected with a convolutional layer for spatial weight prediction. The first LSTM takes the sequential flow density representation as input and generates a hidden state at each time-step for attention map inference, while the second LSTM aims at learning the effective spatial-temporal feature expression from attentionally weighted crowd flow features. Based on the ACFM, we further build a deep architecture with the application to citywide crowd flow prediction, which naturally incorporates the sequential and periodic data as well as other external influences. Extensive experiments on two standard benchmarks (i.e., crowd flow in Beijing and New York City) show that the proposed method achieves significant improvements over the state-of-the-art methods.
Anopheles mating is initiated by the swarming of males at dusk followed by females flying into the swarm. Here, we show that mosquito swarming and mating are coordinately guided by clock genes, light, and temperature. Transcriptome analysis shows up-regulation of the clock genes period (per) and timeless (tim) in the head of field-caught swarming Anopheles coluzzii males. Knockdown of per and tim expression affects Anopheles gambiae s.s. and Anopheles stephensi male mating in the laboratory, and it reduces male An. coluzzii swarming and mating under semifield conditions. Light and temperature affect mosquito mating, possibly by modulating per and/or tim expression. Moreover, the desaturase gene desat1 is up-regulated and rhythmically expressed in the heads of swarming males and regulates the production of cuticular hydrocarbons, including heptacosane, which stimulates mating activity.
Object categories inherently form a hierarchy with di erent levels of concept abstraction, especially for ne-grained categories. For example, birds (Aves) can be categorized according to a four-level hierarchy of order, family, genus, and species. This hierarchy encodes rich correlations among various categories across di erent levels, which can e ectively regularize the semantic space and thus make prediction less ambiguous. However, previous studies of negrained image recognition primarily focus on categories of one certain level and usually overlook this correlation information. In this work, we investigate simultaneously predicting categories of di erent levels in the hierarchy and integrating this structured correlation information into the deep neural network by developing a novel Hierarchical Semantic Embedding (HSE) framework. Specically, the HSE framework sequentially predicts the category score vector of each level in the hierarchy, from highest to lowest. At each level, it incorporates the predicted score vector of the higher level as prior knowledge to learn ner-grained feature representation. During training, the predicted score vector of the higher level is also employed to regularize label prediction by using it as soft targets of corresponding sub-categories. To evaluate the proposed framework, we organize the 200 bird species of the Caltech-UCSD birds dataset with the four-level category hierarchy and construct a large-scale butter y dataset that also covers four level categories. Extensive experiments on these two and the newly-released VegFru datasets demonstrate the superiority of our HSE framework over the baseline methods and existing competitors.
The Grøstl hash function is one of the 5 final round candidates of the SHA-3 competition hosted by NIST. In this paper, we study the preimage resistance of the Grøstl hash function. We propose pseudo preimage attacks on Grøstl hash function for both 256-bit and 512-bit versions, i.e., we need to choose the initial value in order to invert the hash function. Pseudo preimage attack on 5(out of 10)-round Grøstl-256 has a complexity of (2 244.85 , 2 230.13 ) (in time and memory) and pseudo preimage attack on 8(out of 14)-round Grøstl-512 has a complexity of (2 507.32 , 2 507.00 ). To the best of our knowledge, our attacks are the first (pseudo) preimage attacks on round-reduced Grøstl hash function, including its compression function and output transformation. These results are obtained by a variant of meet-in-the-middle preimage attack framework by Aoki and Sasaki. We also improve the time complexities of the preimage attacks against 5-round Whirlpool and 7-round AES hashes by Sasaki in FSE 2011.Grøstl is a double-pipe design, i.e., the size of the chaining value (2n-bit) is twice as the hash size (n-bit). Message length should be less than 2 73 − 577 bits. The padding rule is not introduced here, since it's not important in our attack.The compression function of Grøstl is written as:Where H is the chaining value and M is the message block, both are of 2n bits. After all message blocks are processed, the last chaining value X is used as input of the output transformation, which is written as Ω(X) = T runc n (P (X) ⊕ X)The right half of P (X) ⊕ X is used as the hash value. The compression function and output transformation are illustrated in Fig. 2.
Tubulointerstitial inflammatory cell infiltration and activation contribute to kidney inflammation and fibrosis. Epoxyeicosatrienoic acids (EETs), which are rapidly metabolized to dihydroxyeicosatrienoic acids by the soluble epoxide hydrolase (sEH), have multiple biological functions, including vasodilation, anti-inflammatory action, and others. Inhibition of sEH has been demonstrated to attenuate inflammation in many renal disease models. However, the relationship between sEH expression and macrophage polarization in the kidney remains unknown. In this study, we investigated the relationships between the level of sEH and clinical and pathological parameters in IgA nephropathy. The level of sEH expression positively correlated with proteinuria and infiltration of macrophages. sEH-positive tubules were found to be surrounded by macrophages. Furthermore, we found that incubation of immortalized human proximal tubular HK-2 cells with total urinary protein and overexpression of sEH promoted inflammatory factor production, which was associated with M1 polarization. We also exposed RAW264.7 mouse leukemic monocytes/macrophages to different HK-2 cell culture media conditioned by incubation with various substances affecting sEH amount or activity. We found that the upregulation of sEH promoted M1 polarization. However, pharmacological inhibition of sEH and supplementation with EETs reversed the conditioning effects of urinary proteins by inhibiting M1 polarization through the NF-κB pathway and stimulating M2 polarization through the phosphatidylinositol 3-kinase pathway. These data suggest that inhibition of sEH could be a new strategy to prevent the progression of inflammation and to attenuate renal tubulointerstitial fibrosis.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.