Transition-based top-down parsing with pointer networks has achieved state-of-the-art results in multiple parsing tasks, while having a linear time complexity. However, the decoder of these parsers has a sequential structure, which does not yield the most appropriate inductive bias for deriving tree structures. In this paper, we propose hierarchical pointer network parsers, and apply them to dependency and sentence-level discourse parsing tasks. Our results on standard benchmark datasets demonstrate the effectiveness of our approach, outperforming existing methods and setting a new state-of-the-art.

Generative Adversarial Networks have been crucial in the developments made in unsupervised learning in recent times. Exemplars of image synthesis from text or other images, these networks have shown remarkable improvements over conventional methods in terms of performance. Trained on the adversarial training philosophy, these networks aim to estimate the potential distribution from the real data and then use this as input to generate the synthetic data. Based on this fundamental principle, several frameworks can be generated that are paragon implementations in several real-life applications such as art synthesis, generation of high resolution outputs and synthesis of images from human drawn sketches, to name a few. While theoretically GANs present better results and prove to be an improvement over conventional methods in many factors, the implementation of these frameworks for dedicated applications remains a challenge. This study explores and presents a taxonomy of these frameworks and their use in various image to image synthesis and text to image synthesis applications. The basic GANs, as well as a variety of different niche frameworks, are critically analyzed. The advantages of GANs for image generation over conventional methods as well their disadvantages amongst other frameworks are presented. The future applications of GANs in industries such as healthcare, art and entertainment are also discussed.

Stochastic AUC maximization has garnered an increasing interest due to better fit to imbalanced data classification. However, existing works are limited to stochastic AUC maximization with a linear predictive model, which restricts its predictive power when dealing with extremely complex data. In this paper, we consider stochastic AUC maximization problem with a deep neural network as the predictive model. Building on the saddle point reformulation of a surrogated loss of AUC, the problem can be cast into a {\it non-convex concave} min-max problem. The main contribution made in this paper is to make stochastic AUC maximization more practical for deep neural networks and big data with theoretical insights as well. In particular, we propose to explore Polyak-\L{}ojasiewicz (PL) condition that has been proved and observed in deep learning, which enables us to develop new stochastic algorithms with even faster convergence rate and more practical step size scheme. An AdaGrad-style algorithm is also analyzed under the PL condition with adaptive convergence rate. Our experimental results demonstrate the effectiveness of the proposed algorithms.

Neural network models have been very successful at achieving high accuracy on natural language inference (NLI) tasks. However, as demonstrated in recent literature, when tested on some simple adversarial examples, most of the models suffer a significant drop in performance. This raises the concern about the robustness of NLI models. In this paper, we propose to make NLI models robust by incorporating external knowledge to the attention mechanism using a simple transformation. We apply the new attention to two popular types of NLI models: one is Transformer encoder, and the other is a decomposable model, and show that our method can significantly improve their robustness. Moreover, when combined with BERT pretraining, our method achieves the human-level performance on the adversarial SNLI data set.

A versatile and effective approach to meta-learning is to infer a gradient-based up-date rule directly from data that promotes rapid learning of new tasks from the same distribution. Current methods rely on backpropagating through the learning process, limiting their scope to few-shot learning. In this work, we introduce Warped Gradient Descent (WarpGrad), a family of modular optimisers that can scale to arbitrary adaptation processes. WarpGrad methods meta-learn to warp task loss surfaces across the joint task-parameter distribution to facilitate gradient descent, which is achieved by a reparametrisation of neural networks that interleaves warp layers in the architecture. These layers are shared across task learners and fixed during adaptation; they represent a projection of task parameters into a meta-learned space that is conducive to task adaptation and standard backpropagation induces a form of gradient preconditioning. WarpGrad methods are computationally efficient and easy to implement as they rely on parameter sharing and backpropagation. They are readily combined with other meta-learners and can scale both in terms of model size and length of adaptation trajectories as meta-learning warp parameters do not require differentiation through task adaptation processes. We show empirically that WarpGrad optimisers meta-learn a warped space where gradient descent is well behaved, with faster convergence and better performance in a variety of settings, including few-shot, standard supervised, continual, and reinforcement learning.

To enhance the expressiveness and representational capacity of recurrent neural networks (RNN), a large body of work has emerged exploring stacked architectures with additional topological modifications like shortcut connections or bidirectionality. However, choosing the best network for a particular problem requires a combinatorial search over architectures and their hyperparameters. In this work, we show that a single-layer RNN can perfectly mimic an arbitrarily deep stacked RNN under specific constraints on its weight matrix and a delay between input and output. This obviates the need to manually select hyperparameters like the number of layers. Additionally, we show that weakening weight constraints while keeping the delay gives rise to partial acausality in the single-layer RNN, much like a bidirectional network. Synthetic experiments confirm that the delayed RNN can mimic bidirectional networks in perfectly solving some acausal tasks, outperforming them in others. Finally, we show that in a challenging language processing task, the delayed RNN performs within 0.3\% of the accuracy of the bidirectional network while reducing computational costs.

Machine learning techniques for road networks hold the potential to facilitate many important transportation applications. Graph Convolutional Networks (GCNs) are neural networks that are capable of leveraging the structure of a road network by utilizing information of, e.g., adjacent road segments. While state-of-the-art GCNs target node classification tasks in social, citation, and biological networks, machine learning tasks in road networks differ substantially from such tasks. In road networks, prediction tasks concern edges representing road segments, and many tasks involve regression. In addition, road networks differ substantially from the networks assumed in the GCN literature in terms of the attribute information available and the network characteristics. Many implicit assumptions of GCNs do therefore not apply. We introduce the notion of Relational Fusion Network (RFN), a novel type of GCN designed specifically for machine learning on road networks. In particular, we propose methods that outperform state-of-the-art GCNs on both a road segment regression task and a road segment classification task by 32-40% and 21-24%, respectively. In addition, we provide experimental evidence of the short-comings of state-of-the-art GCNs in the context of road networks: unlike our method, they cannot effectively leverage the road network structure for road segment classification and fail to outperform a regular multi-layer perceptron.

Convolution Neural Networks (CNN) have recently achieved state-of-the art performance on handwritten Chinese character recognition (HCCR). However, most of CNN models employ the SoftMax activation function and minimize cross entropy loss, which may cause loss of inter-class information. To cope with this problem, we propose to combine cross entropy with similarity ranking function and use it as loss function. The experiments results show that the combination loss functions produce higher accuracy in HCCR. This report briefly reviews cross entropy loss function, a typical similarity ranking function: Euclidean distance, and also propose a new similarity ranking function: Average variance similarity. Experiments are done to compare the performances of a CNN model with three different loss functions. In the end, SoftMax cross entropy with Average variance similarity produce the highest accuracy on handwritten Chinese characters recognition.

End-to-end text-to-speech (TTS) synthesis is a method that directly converts input text to output acoustic features using a single network. A recent advance of end-to-end TTS is due to a key technique called attention mechanisms, and all successful methods proposed so far have been based on soft attention mechanisms. However, although network structures are becoming increasingly complex, end-to-end TTS systems with soft attention mechanisms may still fail to learn and to predict accurate alignment between the input and output. This may be because the soft attention mechanisms are too flexible. Therefore, we propose an approach that has more explicit but natural constraints suitable for speech signals to make alignment learning and prediction of end-to-end TTS systems more robust. The proposed system, with the constrained alignment scheme borrowed from segment-to-segment neural transduction (SSNT), directly calculates the joint probability of acoustic features and alignment given an input text. The alignment is designed to be hard and monotonically increase by considering the speech nature, and it is treated as a latent variable and marginalized during training. During prediction, both the alignment and acoustic features can be generated from the probabilistic distributions. The advantages of our approach are that we can simplify many modules for the soft attention and that we can train the end-to-end TTS model using a single likelihood function. As far as we know, our approach is the first end-to-end TTS without a soft attention mechanism.

Sustainable management of groundwater resources under changing climatic conditions require an application of reliable and accurate predictions of groundwater levels. Mechanistic multi-scale, multi-physics simulation models are often too hard to use for this purpose, especially for groundwater managers who do not have access to the complex compute resources and data. Therefore, we analyzed the applicability and performance of four modern deep learning computational models for predictions of groundwater levels. We compare three methods for optimizing the models' hyperparameters, including two surrogate model-based algorithms and a random sampling method. The models were tested using predictions of the groundwater level in Butte County, California, USA, taking into account the temporal variability of streamflow, precipitation, and ambient temperature. Our numerical study shows that the optimization of the hyperparameters can lead to reasonably accurate performance of all models, but the "simplest" network, namely a multilayer perceptron (MLP) performs overall better for learning and predicting groundwater data than the more advanced long short-term memory or convolutional neural networks in terms of prediction accuracy and time-to-solution, making the MLP a suitable candidate for groundwater prediction.