Search

A deep generative model such as a GAN learns to model a rich set of semantic and physical rules about the target distribution, but up to now, it has been obscure how such rules are encoded in the network, or how a rule could be changed. In this paper, we introduce a new problem setting: manipulation of specific rules encoded by a deep generative model. To address the problem, we propose a formulation in which the desired rule is changed by manipulating a layer of a deep network as a linear associative memory. We derive an algorithm for modifying one entry of the associative memory, and we demonstrate that several interesting structural rules can be located and modified within the layers of state-of-the-art generative models. We present a user interface to enable users to interactively change the rules of a generative model to achieve desired effects, and we show several proof-of-concept applications. Finally, results on multiple datasets demonstrate the advantage of our method against standard fine-tuning methods and edit transfer algorithms., ECCV 2020 (oral). Code at https://github.com/davidbau/rewriting. For videos and demos see https://rewriting.csail.mit.edu/

Investigation of the microbial community in the female reproductive tract with the use of sequencing techniques has revealed that endometrial samples obtained through a transvaginal catheter are dominated by Lactobacillus species. Dysbiotic changes in the endometrial microbiota may be associated with implantation failure or early spontaneous abortion in patients who undergo assisted reproductive technology treatment. Whether or not there is an endometrial microbiota in early pregnancy is unknown. Herein we describe, the human endometrial microbiota in a patient who subsequently had an 8th week spontaneous clinical miscarriage with euploid embryos in the next cycle and, for the first time, during a successful pregnancy in which the endometrial fluid was sampled at 4 weeks of gestation. The microbial profile found on the endometrial sample before the spontaneous abortion had higher bacterial diversity and lower Lactobacillus abundance than the endometrial fluid from the healthy pregnancy. Functional metagenomics detected different Lactobacillus species between the 2 samples. Lactobacillus crispatus was present in the endometrium before the spontaneous abortion, as were other bacteria involved in dysbiosis, which had an unstable functional pattern characterized by transposases and insertion elements. Lactobacillus iners was the most prevalent microbe found in the endometrium during early pregnancy; its presence was associated with defense mechanisms and basal functions. These novel observations prompt future investigations to understand the potential implications of microbiology on healthy and pathologic human pregnancy.

The feasibility of geological carbon storage (GCS) sites depends on their capacity to retain safely CO2CO2 . While deep saline formations and depleted gas/oil reservoirs are good candidates to sequester CO2CO2 , gas/oil reservoirs typically have a limited storage capacity compared to ideal targets ( ∼∼ 1 Mt/year) considered for CO2CO2 disposal (Celia et al. in Water Resour Res 51(9):6846–6892, 2015. doi:10.1002/2015WR017609). In this respect, deep saline aquifers are considered more appropriate formations for GCS, but present the disadvantage of having limited characterization data. In particular, information about the continuity of the overlying sealing formations (caprock) is often sparse if it exists at all. In this work, a study of CO2CO2 leakage is conducted for a candidate GCS site located in the Michigan Basin, whose sealing properties of the caprock are practically unknown. Quantification of uncertainty on CO2CO2 leakage from the storage formation is achieved through a Monte Carlo simulation approach, relying on the use of a computationally efficient semi-analytical leakage model based upon the solution derived by Nordbotten et al. (Environ Sci Technol 43(3):743–749, 2009), which assumes leakage occurs across “passive” wells intersecting caprock layers. A categorical indicator Kriging simulator is developed and implemented to represent the caprock sealing properties and model the permeability uncertainty. Binary fields of caprock permeability are generated and exhibit mostly low permeability, with sparsely-occurring local high permeability areas where brine and CO2CO2 may leak out of the storage formation. In addition, the feasibility of extending the use of the semi-analytical model to large-area leakage pathways is studied. This work advances a methodology for preliminary uncertainty quantification of CO2CO2 leakage at sites of GCS with little or no information on the sealing properties of the caprock. The implemented analysis shows that, for the considered site, CO2CO2 leakage may not be negligible even for relatively low ( ∼∼ 1%) probabilities of finding permeable inclusions in the caprock and highlights the importance of being able to characterize caprock sealing properties over large areas.

From 2006 plans were implemented to create a deep-sea water port linked to an Economic Zone in the coastal Province of Ha Tinh, located in north central Vietnam. The multi-purpose Zone entitled ‘Vung Ang’, was to attract foreign investors, while the port would provide a link to nearby Laos and Thailand. The project obviously had large implications for the administrations at various levels of governance from Hanoi to the coastal communes and villages, but even more serious impacts on the people living in the affected areas. A large area of about 23,000 hectares was to be cleared, affecting the people of 9 communes, in some of which all inhabitants had to leave their houses and homesteads, to be relocated to completely new settlements about 10 miles inland. These tightly knit communities were not too happy with the prospect to leave their homes and land, the burial places of their ancestors, and the long term comforts of community support networks. While initial decision making process started at the highest levels of Vietnam Governance, the implementation of port and industrial park construction and the related relocation policy was delegated to Ha Tinh province, which is consistent with current decentralisation policies in Vietnam. Actual implementation was carried out by the affected District and Commune level officials – with support from the Communist Party led Mass Organisations – who were in charge of the planning and implementation of the relocation process. This entailed a complex and sensitive series of steps to inform affected households, prepare relocation areas and allocate compensation and alternative housing. This paper describes the implementation dynamics of relocation by depicting and assessing the roles of all stakeholders involved, including the impacts - for better or for worse – of the relocated households. It brings out the way local authorities dealt with affected people, including efforts linked to the ideal of grass-roots democracy. Key areas of contestation are uncovered, such as inadequate infrastructure and low compensation rates. The paper has a second objective to assess the degree to which decentralisation in Vietnam has been actually implemented, and how this affects policy making processes such as the Vung Ang port/industrial zone project. The paper concludes that the relocation policy was implemented in a fairly efficient and harmonious way – with a very intensive engagement of the entire provincial administrative machinery, but that it is too early to assess the livelihood opportunities of the relocated households.