Porcine reproductive and respiratory syndrome virus (PRRSV), one of the most economically significant pathogens worldwide, has caused numerous outbreaks during the past 30 years. PRRSV infection causes reproductive failure in sows and respiratory disease in growing and finishing pigs, leading to huge economic losses for the swine industry. This impact has become even more significant with the recent emergence of highly pathogenic PRRSV strains from China, further exacerbating global food security. Since new PRRSV variants are constantly emerging from outbreaks, current strategies for controlling PRRSV have been largely inadequate, even though our understanding of PRRSV virology, evolution and host immune response has been rapidly expanding. Meanwhile, practical experience has revealed numerous safety and efficacy concerns for currently licensed vaccines, such as shedding of modified live virus (MLV), reversion to virulence, recombination between field strains and MLV and failure to elicit protective immunity against heterogeneous virus. Therefore, an effective vaccine against PRRSV infection is urgently needed. Here, we systematically review recent advances in PRRSV vaccine development. Antigenic variations resulting from PRRSV evolution, identification of neutralizing epitopes for heterogeneous isolates, broad neutralizing antibodies against PRRSV, chimeric virus generated by reverse genetics, and novel PRRSV strains with interferon-inducing phenotype will be discussed in detail. Moreover, techniques that could potentially transform current MLV vaccines into a superior vaccine will receive special emphasis, as will new insights for future PRRSV vaccine development. Ultimately, improved PRRSV vaccines may overcome the disadvantages of current vaccines and minimize the PRRS impact to the swine industry.

No abstract

No abstract

The ability to decompose complex multi-object scenes into meaningful abstractions like objects is fundamental to achieve higher-level cognition. Previous approaches for unsupervised object-oriented scene representation learning are either based on spatial-attention or scene-mixture approaches and limited in scalability which is a main obstacle towards modeling real-world scenes. In this paper, we propose a generative latent variable model, called SPACE, that provides a unified probabilistic modeling framework that combines the best of spatial-attention and scene-mixture approaches. SPACE can explicitly provide factorized object representations for foreground objects while also decomposing background segments of complex morphology. Previous models are good at either of these, but not both. SPACE also resolves the scalability problems of previous methods by incorporating parallel spatial-attention and thus is applicable to scenes with a large number of objects without performance degradations. We show through experiments on Atari and 3D-Rooms that SPACE achieves the above properties consistently in comparison to SPAIR, IODINE, and GENESIS. Results of our experiments can be found on our project website: https://sites.google.com/view/space-project-page