Structured weight pruning is a representative model compression technique of DNNs to reduce the storage and computation requirements and accelerate inference. An automatic hyperparameter determination process is necessary due to the large number of flexible hyperparameters. This work proposes AutoSlim, an automatic structured pruning framework with the following key performance improvements: (i) effectively incorporate the combination of structured pruning schemes in the automatic process; (ii) adopt the state-of-art ADMM-based structured weight pruning as the core algorithm, and propose an innovative additional purification step for further weight reduction without accuracy loss; and (iii) develop effective heuristic search method enhanced by experience-based guided search, replacing the prior deep reinforcement learning technique which has underlying incompatibility with the target pruning problem. Extensive experiments on CIFAR-10 and ImageNet datasets demonstrate that AutoSlim is the key to achieve ultra-high pruning rates on the number of weights and FLOPs that cannot be achieved before. As an example, AutoSlim outperforms the prior work on automatic model compression by up to 33$\times$ in pruning rate under the same accuracy. We release all models of this work at anonymous link: http://bit.ly/2VZ63dS.

Recent advances in Reinforcement Learning, grounded on combining classical theoretical results with Deep Learning paradigm, led to breakthroughs in many artificial intelligence tasks and gave birth to Deep Reinforcement Learning (DRL) as a field of research. In this work latest DRL algorithms are reviewed with a focus on their theoretical justification, practical limitations and observed empirical properties.

Scene text magnifier aims to magnify text in natural scene images without recognition. It could help the special groups, who have myopia or dyslexia to better understand the scene. In this paper, we design the scene text magnifier through interacted four CNN-based networks: character erasing, character extraction, character magnify, and image synthesis. The architecture of the networks are extended based on the hourglass encoder-decoders. It inputs the original scene text image and outputs the text magnified image while keeps the background unchange. Intermediately, we can get the side-output results of text erasing and text extraction. The four sub-networks are first trained independently and fine-tuned in end-to-end mode. The training samples for each stage are processed through a flow with original image and text annotation in ICDAR2013 and Flickr dataset as input, and corresponding text erased image, magnified text annotation, and text magnified scene image as output. To evaluate the performance of text magnifier, the Structural Similarity is used to measure the regional changes in each character region. The experimental results demonstrate our method can magnify scene text effectively without effecting the background.

Deep learning models are now used in many different industries, while in certain domains safety is not a critical issue in the medical field it is a huge concern. Not only, we want the models to generalize well but we also want to know the models confidence respect to its decision and which features matter the most. Our team aims to develop a full pipeline in which not only displays the uncertainty of the models decision but also, the saliency map to show which sets of pixels of the input image contribute most to the predictions.

In this report, we review the calculation of entropy-regularised Wasserstein loss introduced by Cuturi and document a practical implementation in PyTorch. Recently the Wasserstein distance has seen new applications in machine learning and deep learning. It commonly replaces the Kullback-Leibler divergence (also often dubbed cross-entropy loss in the Deep Learning context). In contrast to the latter, Wasserstein distances not only consider the values probability distribution or density at any given point, but also incorporating spatial information in terms of the underlying metric regarding these differences. Intuitively, it yields a smaller distance if probability mass moved to a nearby point or region and a larger distance if probability mass moved far away.

Statistical inference over structured instances of dependent variables (e.g., labeled sequences, trees, or general graphs) is a fundamental problem in many areas. Examples include computer vision (Nowozin et al., 2011; Dollár & Zitnick, 2013; Chen et al., 2018), natural language processing (Huang et al., 2015; Hu et al., 2016), and computational biology (Li et al., 2007). In many practical setups (Shin et al., 2015; Rekatsinas et al., 2017; Sa et al., 2019; Heidari et al., 2019b), inference problems involve noisy observations of discrete labels assigned to the nodes and edges of a given structured instance and the goal is to infer a labeling of the vertices that achieves low disagreement rate between the correct ground truth labels Y and the predicted labels Ŷ, i.e., low Hamming error. We refer to this problem as statistical recovery. Our motivation to study the problem of statistical recovery stems from our recent work on data cleaning (Rekatsinas et al., 2017; Sa et al., 2019; Heidari et al., 2019b). This work introduces HoloClean, a state-of-the-art inference engine for data curation that casts data cleaning as a structured prediction problem (Sa et al., 2019): Given a dataset as input, it associates each of its cells with a random variable, and uses logical integrity constraints over this dataset (e.g., key constraints or functional dependencies) to introduce dependencies over these random variables. The labels that each random variable can take are determined by the domain of the attribute associated with the corresponding cell. Since we focus on data cleaning, the input dataset corresponds to a noisy version of the latent, clean dataset. Our goal is to recover the latter.

Machine learning approaches for building task-oriented dialogue systems require large conversational datasets with labels to train on. We are interested in building task-oriented dialogue systems from human-human conversations, which may be available in ample amounts in existing customer care center logs or can be collected from crowd workers. Annotating these datasets can be prohibitively expensive. Recently multiple annotated task-oriented human-machine dialogue datasets have been released, however their annotation schema varies across different collections, even for well-defined categories such as dialogue acts (DAs). We propose a Universal DA schema for task-oriented dialogues and align existing annotated datasets with our schema. Our aim is to train a Universal DA tagger (U-DAT) for task-oriented dialogues and use it for tagging human-human conversations. We investigate multiple datasets, propose manual and automated approaches for aligning the different schema, and present results on a target corpus of human-human dialogues. In unsupervised learning experiments we achieve an F1 score of 54.1% on system turns in human-human dialogues. In a semi-supervised setup, the F1 score increases to 57.7% which would otherwise require at least 1.7K manually annotated turns. For new domains, we show further improvements when unlabeled or labeled target domain data is available.

Simulating images representative of neurodegenerative diseases is important for predicting patient outcomes and for validation of computational models of disease progression. This capability is valuable for secondary prevention clinical trials where outcomes and screening criteria involve neuroimaging. Traditional computational methods are limited by imposing a parametric model for atrophy and are extremely resource-demanding. Recent advances in deep learning have yielded data-driven models for longitudinal studies (e.g., face ageing) that are capable of generating synthetic images in real-time. Similar solutions can be used to model trajectories of atrophy in the brain, although new challenges need to be addressed to ensure accurate disease progression modelling. Here we propose Degenerative Adversarial NeuroImage Net (DaniNet) --- a new deep learning approach that learns to emulate the effect of neurodegeneration on MRI. DaniNet uses an underlying set of Support Vector Regressors (SVRs) trained to capture the patterns of regional intensity changes that accompany disease progression. DaniNet produces whole output images, consisting of 2D-MRI slices that are constrained to match regional predictions from the SVRs. DaniNet is also able to condition the progression on non-imaging characteristics (age, diagnosis, etc.) while it maintains the unique brain morphology of individuals. Adversarial training ensures realistic brain images and smooth temporal progression. We train our model using 9652 T1-weighted (longitudinal) MRI extracted from the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We perform quantitative and qualitative evaluations on a separate test set of 1283 images (also from ADNI) demonstrating the ability of DaniNet to produce accurate and convincing synthetic images that emulate disease progression.

This paper targets the problem of image set-based face verification and identification. Unlike traditional single media (an image or video) setting, we encounter a set of heterogeneous contents containing orderless images and videos. The importance of each image is usually considered either equal or based on their independent quality assessment. How to model the relationship of orderless images within a set remains a challenge. We address this problem by formulating it as a Markov Decision Process (MDP) in the latent space. Specifically, we first present a dependency-aware attention control (DAC) network, which resorts to actor-critic reinforcement learning for sequential attention decision of each image embedding to fully exploit the rich correlation cues among the unordered images. Moreover, we introduce its sample-efficient variant with off-policy experience replay to speed up the learning process. The pose-guided representation scheme can further boost the performance at the extremes of the pose variation.

Attention operators have been widely applied in various fields, including computer vision, natural language processing, and network embedding learning. Attention operators on graph data enables learnable weights when aggregating information from neighboring nodes. However, graph attention operators (GAOs) consume excessive computational resources, preventing their applications on large graphs. In addition, GAOs belong to the family of soft attention, instead of hard attention, which has been shown to yield better performance. In this work, we propose novel hard graph attention operator (hGAO) and channel-wise graph attention operator (cGAO). hGAO uses the hard attention mechanism by attending to only important nodes. Compared to GAO, hGAO improves performance and saves computational cost by only attending to important nodes. To further reduce the requirements on computational resources, we propose the cGAO that performs attention operations along channels. cGAO avoids the dependency on the adjacency matrix, leading to dramatic reductions in computational resource requirements. Experimental results demonstrate that our proposed deep models with the new operators achieve consistently better performance. Comparison results also indicates that hGAO achieves significantly better performance than GAO on both node and graph embedding tasks. Efficiency comparison shows that our cGAO leads to dramatic savings in computational resources, making them applicable to large graphs.