Boosting Weakly Supervised Object Localization and Segmentation With Domain Adaption.

Weakly supervised object localization (WSOL), adopting only image-level annotations to learn the pixel-level localization model, can release human resources in the annotation process. Most one-stage WSOL methods learn the localization model with multi-instance learning, making them only activate discriminative object parts rather than the whole object. In our work, we attribute this problem to the domain shift between the training and test process of WSOL and provide a novel perspective that views WSOL as a domain adaption (DA) task. Under this perspective, a DA-WSOL pipeline is elaborated to better assist WSOL with DA approaches by considering the specificities for the adaption of WSOL. Our DA-WSOL pipeline can discern the source-related and the Universum samples from other target samples based on a proposed target sampling strategy and then utilize them to solve the sample unbalancing and label unmatching between the source and target domain of WSOL. Experiments show that our pipeline outperforms SOTA methods on three WSOL benchmarks and can improve the performance of downstream weakly supervised semantic segmentation tasks. Codes are available at https://github.com/zh460045050/dawsol.

The association of SOD and HsCRP with the efficacy of sulforaphane in schizophrenia patients with residual negative symptoms.

OBJECTIVES: Emerging evidence indicates a connection between oxidative stress, immune-inflammatory processes, and the negative symptoms of schizophrenia. In addition to possessing potent antioxidant and anti-inflammatory properties, sulforaphane (SFN) has shown promise in enhancing cognitive function among individuals with schizophrenia. This study aims to investigate the efficacy of combined treatment with SFN in patients with schizophrenia who experience negative symptoms and its effect on the levels of superoxide dismutase (SOD) and the inflammatory marker, high-sensitivity C-reactive protein (HsCRP). DESIGN: Forty-five patients with schizophrenia were recruited, who mainly experienced negative symptoms during a stable period. In addition to the original treatments, the patients received SFN tablets at a daily dose of 90 mg for 24 weeks. At baseline, 12 weeks, and 24 weeks, the participants were interviewed and evaluated. The reduction rate of the Positive and Negative Syndrome Scale (PANSS) was used to assess each participant. The side effects scale of Treatment Emergent Symptom Scale (TESS) was applied to assess the adverse reactions. Additionally, the levels of the SOD, HsCRP, and other indicators were examined. RESULTS: The study findings revealed a significant decrease in PANSS negative subscale scores (P < 0.001). Furthermore, there was a significant increase in SOD activity and HsCRP levels (P < 0.001 and P < 0.05). Notably, the group of participants who exhibited a reduction in PANSS negative subscale scores demonstrated a significant improvement in HsCRP levels (P < 0.05). CONCLUSIONS: Our study suggests that SFN may potentially serve as a safe adjunctive intervention to improve the negative symptoms of schizophrenia. The potential mechanism by which SFN improves negative symptoms in schizophrenia patients may involve its anti-inflammatory properties, specifically its ability to reduce HsCRP levels. Trial registration ClinicalTrial.gov (ID: NCT03451734).

Background-Aware Classification Activation Map for Weakly Supervised Object Localization.

Weakly supervised object localization (WSOL) relaxes the requirement of dense annotations for object localization by using image-level annotation to supervise the learning process. However, most WSOL methods only focus on forcing the object classifier to produce high activation score on object parts without considering the influence of background locations, causing excessive background activations and ill-pose background score searching. Based on this point, our work proposes a novel mechanism called the background-aware classification activation map (B-CAM) to add background awareness for WSOL training. Besides aggregating an object image-level feature for supervision, our B-CAM produces an additional background image-level feature to represent the pure-background sample. This additional feature can provide background cues for the object classifier to suppress the background activations on object localization maps. Moreover, our B-CAM also trained a background classifier with image-level annotation to produce adaptive background scores when determining the binary localization mask. Experiments indicate the effectiveness of the proposed B-CAM on four different types of WSOL benchmarks, including CUB-200, ILSVRC, OpenImages, and VOC2012 datasets.

Comparative transcriptomics provides insights into the pathogenic immune response of brown leaf spots in weeping forsythia.

Weeping forsythia is an important ornamental, ecological and medicinal plant. Brown leaf spots limit the large-scale production of weeping forsythia as a medicinal crop. Alternaria alternata is a pathogen causing brown leaf spots in weeping forsythia; however, its pathogenesis and the immune response mechanisms of weeping forsythia remain unclear. In this study, we identified two mechanisms based on morphological anatomy, physiological indexes and gene expression analyses. Our results showed that A. alternata induced leaf stomata to open, invaded the mesophyll, dissolved the cell wall, destroyed the cell membrane and decreased the number of chloroplasts by up-regulating the expression of auxin-activated signaling pathway genes. Alternaria alternata also down-regulated iron-ion homeostasis and binding-related genes, which caused an increase in the levels of iron ions and reactive oxygen species in leaves. These processes eventually led to programmed cell death, destroying palisade and spongy tissues and causing the formation of iron rust spots. Alternaria alternata also caused defense and hypersensitive responses in weeping forsythia through signaling pathways mediated by flg22-like and elf18-like polypeptides, ethylene, H2O2 and bacterial secretion systems. Our study provides a theoretical basis for the control of brown leaf spots in weeping forsythia.

Power Law in Deep Neural Networks: Sparse Network Generation and Continual Learning With Preferential Attachment.

Training deep neural networks (DNNs) typically requires massive computational power. Existing DNNs exhibit low time and storage efficiency due to the high degree of redundancy. In contrast to most existing DNNs, biological and social networks with vast numbers of connections are highly efficient and exhibit scale-free properties indicative of the power law distribution, which can be originated by preferential attachment in growing networks. In this work, we ask whether the topology of the best performing DNNs shows the power law similar to biological and social networks and how to use the power law topology to construct well-performing and compact DNNs. We first find that the connectivities of sparse DNNs can be modeled by truncated power law distribution, which is one of the variations of the power law. The comparison of different DNNs reveals that the best performing networks correlated highly with the power law distribution. We further model the preferential attachment in DNNs evolution and find that continual learning in networks with growth in tasks correlates with the process of preferential attachment. These identified power law dynamics in DNNs can lead to the construction of highly accurate and compact DNNs based on preferential attachment. Inspired by the discovered findings, two novel applications have been proposed, including evolving optimal DNNs in sparse network generation and continual learning tasks with efficient network growth using power law dynamics. Experimental results indicate that the proposed applications can speed up training, save storage, and learn with fewer samples than other well-established baselines. Our demonstration of preferential attachment and power law in well-performing DNNs offers insight into designing and constructing more efficient deep learning.

Search for clinical predictors of good glycemic control in patients starting or intensifying oral hypoglycemic pharmacological therapy: A multicenter prospective cohort study.

AIMS: Our aim was to search for clinical predictors of good glycemic control in patients starting or intensifying oral hypoglycemic pharmacological therapy. METHODS: A multicenter, prospective cohort of 499 diabetic subjects was enrolled in this study: patients with newly diagnosed diabetes (NDM group) or poor glycemic control with oral antidiabetic drugs (OADs) (PDM group). All subjects then started or intensified OADs therapy and followed up for 91 days. Glycemic control was determined according to HbA1c at day 91 with HbA1c <7% considered good. RESULTS: The proportions of patients with good glycemic control after follow up for 91 days were 66.9% and 34.8% in NDM group and PDM group respectively. Logistic regression analysis showed that the change in GA at 28 days was the only predictor of good glycemic control in NDM patients (OR = 1.630, 95% CI 1.300-2.044, P < 0.001). In PDM patients, changes in GA at 28 days, CPI, baseline HbA1c, diabetic duration, and BMI were all independent predictors of good glycemic control (All P < 0.05). CONCLUSIONS: GA decline is a good predictor of future success in newly diagnosed patients. In patients intensifying therapy, beside GA decline, other individualized clinical characteristics should also be considered.

Chemical composition, antimicrobial activity and antioxidant activity of Litsea cubeba essential oils in different months.

In this study, the chemical composition, antimicrobial and antioxidant activities of Litsea cubeba essential oils extracted in different months were analysed. Results showed that the essential oil contents of fruits collected in June, July and August were 3.47%, 5.02% and 5.64%, respectively, and contained 13, 17 and 17 components, respectively. Neral and geranial were the main components and accounted for 54.76%. The essential oil extracted from fruits collected in July had the highest antimicrobial activity against Staphylococcus aureus, Escherichia coli, and Salmonella typhimurium, and it was the most effective based on the OH· scavenging activity test. The essential oil extracted from fruits collected in August was the most effective based on the test for DPPH· scavenging activity and ferric reducing antioxidant power. Considering the contents, chemical compositions and antimicrobial and antioxidant activities, the appropriate harvest time for L. cubeba essential oils is from July to August.

Weighted Network Density Predicts Range of Latent Variable Model Accuracy.

Current experimental techniques impose spatial limits on the number of neuronal units that can be recorded invivo. To model the neuronal dynamics utilizing these sampled data, Latent Variable Models (LVMs) have been proposed to study the common unobserved processes within the system that drives neuronal activities, through an implicit network with hidden states. Yet, relationships between these latent variable models and widely-studied network connectivity measures have remained unclear. In this paper, a biologically plausible latent variable model was fit to neuronal activity recorded via 2-photon microscopic calcium imaging in the murine primary visual cortex. Graph theoretic measures were then applied to quantify network properties in the recorded sub-regions. Comparison of weighted network measures with LVM prediction accuracy shows some network measures having a strong relationship with LVM prediction accuracy, while other measures do not have a robust relationship with LVM prediction accuracy. Results show LVM will achieve high accuracy in dense networks.

Glycated albumin is superior to glycated hemoglobin for glycemic control assessment at an early stage of diabetes treatment: A multicenter, prospective study.

AIMS: This study was to determine whether serum glycated albumin (GA) was a better indicator of glycemic control than hemoglobin A1c (HbA1c) when starting a new treatment regimen for type 2 diabetes. METHODS: Newly diagnosed type 2 diabetes patients, or patients who had poor glycemic control with oral hypoglycemic agents, were enrolled at 10 hospitals in Beijing. Serum GA, HbA1c, fasting blood glucose (FBG), and C-peptide were assayed on Days 0, 14, 28, and 91 after treatment. RESULTS: Four hundred ninety-nine patients were enrolled. Mean FBG, GA and HbA1c decreased significantly in patients at Days 14, 28, and 91. In patients with improved glycemic control, the reduction of GA and HbA1c levels was 10.5±13.3% vs. 5.1±5.4% on Day 14, 16.0±13.4% vs. 9.0±7.0% on Day 28, and 18.0±16.7% vs. 18.3±9.4% on Day 91, respectively, compared with baseline values. Changes in GA on Day 14, 28 and 91 were all closely correlated with changes in HbA1c on Day 91. Change in GA on Day 14 was correlated with treatment effectiveness evaluated by HbA1c on Day 91. CONCLUSIONS: GA may be a useful marker for assessing glycemic control at an early stage of new diabetes treatment and assist in guiding adjustments to treatment and therapy.

Evaluating the Small-World-Ness of a Sampled Network: Functional Connectivity of Entorhinal-Hippocampal Circuitry.

The amount of publicly accessible experimental data has gradually increased in recent years, which makes it possible to reconsider many longstanding questions in neuroscience. In this paper, an efficient framework is presented for reconstructing functional connectivity using experimental spike-train data. A modified generalized linear model (GLM) with L1-norm penalty was used to investigate 10 datasets. These datasets contain spike-train data collected from the entorhinal-hippocampal region in the brains of rats performing different tasks. The analysis shows that entorhinal-hippocampal network of well-trained rats demonstrated significant small-world features. It is found that the connectivity structure generated by distance-dependent models is responsible for the observed small-world features of the reconstructed networks. The models are utilized to simulate a subset of units recorded from a large biological neural network using multiple electrodes. Two metrics for quantifying the small-world-ness both suggest that the reconstructed network from the sampled nodes estimates a more prominent small-world-ness feature than that of the original unknown network when the number of recorded neurons is small. Finally, this study shows that it is feasible to adjust the estimated small-world-ness results based on the number of neurons recorded to provide a more accurate reference of the network property.

Reconstruction of neural network topology using spike train data: Small-world features of hippocampal network.

As the amount of experimental data made publicly accessible has gradually increased in recent years, it is now possible to reconsider many of the longstanding questions in neuroscience. In this paper, we present an efficient frame-work for reconstructing the functional connectivity from the spike train data curated from the Collaborative Research in Computational Neuroscience (CRCNS) program. We used a modified generalized linear model (GLM) framework with L1 norm penalty to investigate 10 datasets. These datasets contain spike train data collected from the hippocampal region of rats performing various tasks. Analysis of the reconstructed network showed that the neural network in the hippocampal region of well-trained rats demonstrated significant small-world features.

Effect of dexamethasone on peroxisome proliferator activated receptor-gamma mRNA expression in 3T3-L1 adipocytes with the human recombinant adiponectin.

BACKGROUND: The fat derived protein adiponectin plays an important role in the regulation of glucose metabolism. The aim of this study was to provide the experimental basis for further investigating on adiponectin (ADPN) function. Its eukaryotic recombinant was constructed and expressed in precursor cells of 3T3-L1 adipocytes. The effects of dexamethasone on peroxisome proliferator activated receptor-gamma (PPAR-gamma) mRNA expression in 3T3-L1 cells with human recombinant adiponectin were assessed. METHODS: The recombinant plasmid pMD18-T-hADPN and eukaryotic expression vector pcDNA3.1(+) were digested by two restrictive endonucleases and adiponectin and linear pcDNA3.1(+) were obtained. Then, they were ligated and translated into JM109. The recombinant pcDNA3.1(+)-hADPN so obtained was identified by digestion by restrictive endonuclease and nucleotide sequencing. The 3T3-L1 precursor cells were transfected using SuperFect Transfection Reagent (Qiagen). Furthermore, 3T3-L1 cells with human recombinant adiponectin incubated with dexamethasone (0.5 mmol/L) for 24 hours, cells were collected and total RNA was extracted. The PPAR-gamma mRNA expression was quantified by semiquantitative reverse transcription-polymerase chain reaction (RT-PCR). RESULTS: After eukaryotic recombinant was digested by Hind III and EcoR I, fragments of 800 bp and 5.4 kb were identified by nucleotide sequence scanning and consistent with theoretical values. Electrophoretogram of RT-PCR in 3T3-L1 precursors showed only one band in front of 250 bp, which was consistent with theoretical value 234 bp. In the 3T3-L1 cells, 3T3-L1 cells with plasmid and 3T3-L1 cells human recombinant adiponectin, treatment with dexamethasone (0.5 mmol/L) decreased PPAR-gamma mRNA expression compared to untreated controls (P < 0.01). Effect of dexamethasone on PPAR-gamma mRNA expression in 3T3-L1 cells was reversed by stably transfected human recombinant adiponectin. CONCLUSION: The 3T3-L1 cells stably transfected human recombinant adiponectin had increased PPAR-gamma mRNA expression. Dexamethasone suppressed PPAR-gamma mRNA expression in the 3T3-L1 cells. Effect of dexamethasone on PPAR-gamma mRNA expression in 3T3-L1 cells was reversed by stably transfected human recombinant adiponectin.