We propose a novel multi-task learning architecture, which allows learning of task-specific feature-level attention. Our design, the Multi-Task Attention Network (MTAN), consists of a single shared network containing a global feature pool, together with a soft-attention module for each task. These modules allow for learning of taskspecific features from the global features, whilst simultaneously allowing for features to be shared across different tasks. The architecture can be trained end-to-end and can be built upon any feed-forward neural network, is simple to implement, and is parameter efficient. We evaluate our approach on a variety of datasets, across both image-toimage predictions and image classification tasks. We show that our architecture is state-of-the-art in multi-task learning compared to existing methods, and is also less sensitive to various weighting schemes in the multi-task loss function. Code is available at https://github.com/ lorenmt/mtan.

Recent evidence shows that resveratrol (RSV) may ameliorate high-glucose-induced cardiac oxidative stress, mitochondrial dysfunction and myocardial fibrosis in diabetes. However, the mechanisms by which RSV regulates mitochondrial function in diabetic cardiomyopathy have not been fully elucidated. Mitochondrial dysfunction contributes to cardiac dysfunction in diabetic patients, which is associated with dysregulation of peroxisome proliferator-activated receptor gamma coactivator-1α (PGC-1α). In this study we examined whether resveratrol alleviated cardiac dysfunction in diabetes by improving mitochondrial function via SIRT1-mediated PGC-1α deacetylation. T2DM was induced in rats by a high-fat diet combined with STZ injection. Diabetic rats were orally administered RSV (50 mg·kg·d) for 16 weeks. RSV administration significantly attenuated diabetes-induced cardiac dysfunction and hypertrophy evidenced by increasing ejection fraction (EF%), fraction shortening (FS%), ratio of early diastolic peak velocity (E velocity) and late diastolic peak velocity (A velocity) of the LV inflow (E/A ratio) and reducing expression levels of pro-hypertrophic markers ANP, BNP and β-MHC. Furthermore, manganese superoxide dismutase (SOD) activity, ATP content, mitochondrial DNA copy number, mitochondrial membrane potential and the expression of nuclear respiration factor (NRF) were all significantly increased in diabetic hearts by RSV administration, whereas the levels of malondialdehvde (MDA) and uncoupling protein 2 (UCP2) were significantly decreased. Moreover, RSV administration significantly activated SIRT1 expression and increased PGC-1α deacetylation. H9c2 cells cultured in a high glucose (HG, 30 mmol/L) condition were used for further analyzing the role of SIRT1/PGC-1α pathway in RSV regulation of mitochondrial function. RSV (20 μmol/L) caused similar beneficial effects in HG-treated H9c2 cells in vitro as in diabetic rats, but these protective effects were abolished by addition of a SIRT1 inhibitor sirtinol (25 μmol/L) or by SIRT1 siRNA transfection. In H9c2 cells, RSV-induced PGC-1α deacetylation was dependent on SIRT1, which was also abolished by a SIRT1 inhibitor and SIRT1 siRNA transfection. Our results demonstrate that resveratrol attenuates cardiac injury in diabetic rats through regulation of mitochondrial function, which is mediated partly through SIRT1 activation and increased PGC-1α deacetylation.

We show for the first time that a multilayer perceptron (MLP) can serve as the only scene representation in a realtime SLAM system for a handheld RGB-D camera. Our network is trained in live operation without prior data, building a dense, scene-specific implicit 3D model of occupancy and colour which is also immediately used for tracking.Achieving real-time SLAM via continual training of a neural network against a live image stream requires significant innovation. Our iMAP algorithm uses a keyframe structure and multi-processing computation flow, with dynamic information-guided pixel sampling for speed, with tracking at 10 Hz and global map updating at 2 Hz. The advantages of an implicit MLP over standard dense SLAM techniques include efficient geometry representation with automatic detail control and smooth, plausible filling-in of unobserved regions such as the back surfaces of objects.

No abstract