In this paper, we address referring expression comprehension: localizing an image region described by a natural language expression. While most recent work treats expressions as a single unit, we ...propose to decompose them into three modular components related to subject appearance, location, and relationship to other objects. This allows us to flexibly adapt to expressions containing different types of information in an end-to-end framework. In our model, which we call the Modular Attention Network (MAttNet), two types of attention are utilized: language-based attention that learns the module weights as well as the word/phrase attention that each module should focus on; and visual attention that allows the subject and relationship modules to focus on relevant image components. Module weights combine scores from all three modules dynamically to output an overall score. Experiments show that MAttNet outperforms previous state-of-the-art methods by a large margin on both bounding-box-level and pixel-level comprehension tasks. Demo1 and code2 are provided.
Recently, fine-tuning language models pre-trained on large text corpora have provided huge improvements on vision-and-language (V&L) tasks as well as on pure language tasks. However, fine-tuning the ...entire parameter set of pre-trained models becomes impractical since the model size is growing rapidly. Hence, in this paper, we introduce adapter-based parameter-efficient transfer learning techniques to V&L models such as VL-BART and VL-T5. We evaluate our methods in a unified multi-task setup on both image-text and video-text benchmarks. For the image-text tasks, we use four diverse V&L datasets: VQAv2, GQA, NLVR 2 , and MSCOCO image captioning. For video-text tasks, we use TVQA, How2QA, TVC, and YC2C. With careful training and thorough experiments, we benchmark three popular adapter-based methods (Adapter, Hyperformer, Compacter) against the standard full fine-tuning and the recently proposed prompt-tuning approach. We also enhance the efficiency and performance of adapters by sharing their weights to attain knowledge across tasks. Our results demonstrate that training the adapter with the weight-sharing technique (4.18% of total parameters for image-text tasks and 3.39% for video-text tasks) can match the performance of fine-tuning the entire model. Lastly, we present a comprehensive analysis including the combination of adapter and task-specific prompts and the impact of V&L pre-training on adapters. 1 1 The code for our CVPR 2022 paper is available at: https://github.com/ylsung/VL_adapter.
NLP has achieved great progress in the past decade through the use of neural models and large labeled datasets. The dependence on abundant data prevents NLP models from being applied to low-resource ...settings or novel tasks where significant time, money, or expertise is required to label massive amounts of textual data. Recently, data augmentation methods have been explored as a means of improving data efficiency in NLP. To date, there has been no
of data augmentation for NLP in the limited labeled data setting, making it difficult to understand which methods work in which settings. In this paper, we provide an
of recent progress on data augmentation for NLP in the limited labeled data setting, summarizing the landscape of methods (including token-level augmentations, sentence-level augmentations, adversarial augmentations, and hidden-space augmentations) and carrying out experiments on 11 datasets covering topics/news classification, inference tasks, paraphrasing tasks, and single-sentence tasks. Based on the results, we draw several conclusions to help practitioners choose appropriate augmentations in different settings and discuss the current challenges and future directions for limited data learning in NLP.
Salmonella Enteritidis is an intracellular foodborne pathogen that has developed multiple mechanisms to alter poultry intestinal physiology and infect the gut. Short chain fatty acid butyrate is ...derived from microbiota metabolic activities, and it maintains gut homeostasis. There is limited understanding on the interaction between S. Enteritidis infection, butyrate, and host intestinal response. To fill this knowledge gap, chicken macrophages (also known as HTC cells) were infected with S. Enteritidis, treated with sodium butyrate, and proteomic analysis was performed. A growth curve assay was conducted to determine sub-inhibitory concentration (SIC, concentration that do not affect bacterial growth compared to control) of sodium butyrate against S. Enteritidis. HTC cells were infected with S. Enteritidis in the presence and absence of SIC of sodium butyrate. The proteins were extracted and analyzed by tandem mass spectrometry. Our results showed that the SIC was 45 mM. Notably, S. Enteritidis-infected HTC cells upregulated macrophage proteins involved in ATP synthesis through oxidative phosphorylation such as ATP synthase subunit alpha (ATP5A1), ATP synthase subunit d, mitochondrial (ATP5PD) and cellular apoptosis such as Cytochrome-c (CYC). Furthermore, sodium butyrate influenced S. Enteritidis-infected HTC cells by reducing the expression of macrophage proteins mediating actin cytoskeletal rearrangements such as WD repeat-containing protein-1 (WDR1), Alpha actinin-1 (ACTN1), Vinculin (VCL) and Protein disulfide isomerase (P4HB) and intracellular S. Enteritidis growth and replication such as V-type proton ATPase catalytic subunit A (ATPV1A). Interestingly, sodium butyrate increased the expression of infected HTC cell protein involving in bacterial killing such as Vimentin (VIM). In conclusion, sodium butyrate modulates the expression of HTC cell proteins essential for S. Enteritidis invasion.
Neuroblastoma is a pediatric cancer responsible for approximately 15% of all childhood cancer deaths. Aberrant MYCN activation, as a result of genomic MYCN amplification, is a major driver of ...high-risk neuroblastoma, which has an overall survival rate of less than 50%, despite the best treatments currently available. Metabolic reprogramming is an integral part of the growth-promoting program driven by MYCN, which fuels cell growth and proliferation by increasing the uptake and catabolism of nutrients, biosynthesis of macromolecules, and production of energy. This reprogramming process also generates metabolic vulnerabilities that can be exploited for therapy. In this review, we present our current understanding of metabolic reprogramming in neuroblastoma, focusing on transcriptional regulation as a key mechanism in driving the reprogramming process. We also highlight some important areas that need to be explored for the successful development of metabolism-based therapy against high-risk neuroblastoma.
Despite reducing the prevalent foodborne pathogen Campylobacter jejuni in chickens decreases campylobacteriosis, few effective approaches are available. The aim of this study was to use microbial ...metabolic product bile acids to reduce C. jejuni chicken colonization. Broiler chicks were fed with deoxycholic acid (DCA), lithocholic acid (LCA), or ursodeoxycholic acid (UDCA). The birds were also transplanted with DCA modulated anaerobes (DCA-Anaero) or aerobes (DCA-Aero). The birds were infected with human clinical isolate C. jejuni 81-176 or chicken isolate C. jejuni AR101. Notably, C. jejuni 81-176 was readily colonized intestinal tract at d16 and reached an almost plateau at d21. Remarkably, DCA excluded C. jejuni cecal colonization below the limit of detection at 16 and 28 days of age. Neither chicken ages of infection nor LCA or UDCA altered C. jejuni AR101 chicken colonization level, while DCA reduced 91% of the bacterium in chickens at d28. Notably, DCA diet reduced phylum Firmicutes but increased Bacteroidetes compared to infected control birds. Importantly, DCA-Anaero attenuated 93% of C. jejuni colonization at d28 compared to control infected birds. In conclusion, DCA shapes microbiota composition against C. jejuni colonization in chickens, suggesting a bidirectional interaction between microbiota and microbial metabolites.
•The concept of “Effective Hamiltonians” is employed to describe finite pulse effects in spin I=1/2 systems.•The role of interaction frames in the exactness of the effective Hamiltonians is discussed ...along with the convergence criterion.•Depending on the magnitudes of the internal and external parameters, the analytic methods are classified into different regimes.•The limitations of existing analytic methods are also discussed.
Display omitted
In this mini-review, we examine the exactness of the available theoretical methods for describing the effects of a radio frequency (RF) pulse on a spin I=1/2 system. Employing an isolated spin as a model system, the interplay between the internal (anisotropic) interactions and the external parameters (such as sample spinning frequency, amplitude of the pulse etc.) is discussed through comparison between simulations emerging from analytic and numerical methods. Accordingly, the suitability of the analytic methods are examined and classified into regimes based on the magnitudes of the internal and external parameters.
Campylobacter jejuni is a prevalent foodborne pathogen mainly transmitting through poultry. It remains unknown how chicken-transmitted C. jejuni and microbiota impact on human campylobacteriosis. ...Campylobacter jejuni AR101 (Cj-P0) was introduced to chickens and isolated as passage 1 (Cj-P1). Campylobacter jejuni Cj-P1-DCA-Anaero was isolated from Cj-P0-infected birds transplanted with DCA-modulated anaerobic microbiota. Specific pathogen free Il10
mice were gavaged with antibiotic clindamycin and then infected with Cj-P0, Cj-P1, or Cj-P1-DCA-Anaero, respectively. After 8 days post infection, Il10
mice infected with Cj-P1 demonstrated severe morbidity and bloody diarrhea and the experiment had to be terminated. Cj-P1 induced more severe histopathology compared to Cj-P0, suggesting that chicken transmission increased C. jejuni virulence. Importantly, mice infected with Cj-P1-DCA-Anaero showed attenuation of intestinal inflammation compared to Cj-P1. At the cellular level, Cj-P1 induced more C. jejuni invasion and neutrophil infiltration into the Il10
mouse colon tissue compared to Cj-P0, which was attenuated with Cj-P1-DCA-Anaero. At the molecular level, Cj-P1 induced elevated inflammatory mediator mRNA accumulation of Il17a, Il1β, and Cxcl1 in the colon compared to Cj-P0, while Cj-P1-DCA-Anaero showed reduction of the inflammatory gene expression. In conclusion, our data suggest that DCA-modulated anaerobes attenuate chicken-transmitted campylobacteriosis in mice and it is important to control the elevation of C. jejuni virulence during chicken transmission process.
The objective of this study was to determine the effect of chlorine induced sublethal oxidative stress against homologous and heterologous stress adaptations in five
(
) strains.
cells were exposed ...to gradually increasing sublethal concentrations of total chlorine/day: 250 ppm (day 1), 270 ppm (day 2), 290 ppm (day 3), 310 ppm (day 4), 330 ppm (day 5), 350 ppm (day 6), and 375 ppm (day 7) in tryptic soy broth (TSB). Changes in minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) of
cells exposed to chlorine and control (non-adapted cells) were determined by the macro-dilution method. Chlorine-adapted
cells were also evaluated for changes in antibiotic resistance using the Kirby-Bauer disk diffusion and MIC double dilution assay as per the Clinical and Laboratory Standards Institute (CLSI, 2016) guidelines. In four
strains (Scott A, V7, FSL-N1-227 and FSL-F6-154) after adapted to sublethal chlorine, the MIC (600 ppm) and MBC (700 ppm) values of chlorine were slightly higher as compared to control (500 ppm MIC, and 600 ppm MBC). The Kirby-Bauer and MIC double dilution assays showed some significant changes in antibiotic susceptibility patterns for antibiotics such as streptomycin, gentamicin and ceftriaxone (
< 0.05). However, the changes in zones of inhibition and MIC values to all antibiotics tested for the chlorine-adapted and non-adapted (control)
cells were still within the susceptible range. Transmission electron microscopy studies showed that changes in cell wall and membrane integrity resulting, from the elongation of cells, may contribute to the possible routes of its increase in tolerance to chlorine and selective antibiotics. These findings indicate that the continuous exposure of
cells to chlorine may lead to significant changes in homologs and heterologous stress adaptation.