Recent studies illustrate how machine learning (ML) can be used to bypass a core challenge of molecular modeling: the tradeoff between accuracy and computational cost. Here, we assess multiple ML approaches for predicting the atomization energy of organic molecules. Our resulting models learn the difference between low-fidelity, B3LYP, and high-accuracy, G4MP2, atomization energies, and predict the G4MP2 atomization energy to 0.005 eV (mean absolute error) for molecules with less than 9 heavy atoms and 0.012 eV for a small set of molecules with between 10 and 14 heavy atoms. Our two best models, which have different accuracy/speed tradeoffs, enable the efficient prediction of G4MP2-level energies for large molecules and are available through a simple web interface.

Low-precision representation of deep neural networks (DNNs) is critical for efficient deployment of deep learning application on embedded platforms, however, converting the network to low precision degrades its performance. Crucially, networks that are designed for embedded applications usually suffer from increased degradation since they have less redundancy. This is most evident for the ubiquitous MobileNet architecture which requires a costly quantization-aware training cycle to achieve acceptable performance when quantized to 8-bits. In this paper, we trace the source of the degradation in MobileNets to a shift in the mean activation value. This shift is caused by an inherent bias in the quantization process which builds up across layers, shifting all network statistics away from the learned distribution. We show that this phenomenon happens in other architectures as well. We propose a simple remedy - compensating for the quantization induced shift by adding a constant to the additive bias term of each channel. We develop two simple methods for estimating the correction constants - one using iterative evaluation of the quantized network and one where the constants are set using a short training phase. Both methods are fast and require only a small amount of unlabeled data, making them appealing for rapid deployment of neural networks. Using the above methods we are able to match the performance of training-based quantization of MobileNets at a fraction of the cost.

Capsule Networks attempt to represent patterns in images in a way that preserves hierarchical spatial relationships. Additionally, research has demonstrated that these techniques may be robust against adversarial perturbations. We present an improvement to training capsule networks with added robustness via non-parametric kernel methods. The representations learned through the capsule network are used to construct covariance kernels for Gaussian processes (GPs). We demonstrate that this approach achieves comparable prediction performance to Capsule Networks while improving robustness to adversarial perturbations and providing a meaningful measure of uncertainty that may aid in the detection of adversarial inputs.

Deep Neural Networks have shown tremendous success in the area of object recognition, image classification and natural language processing. However, designing optimal Neural Network architectures that can learn and output arbitrary graphs is an ongoing research problem. The objective of this survey is to summarise and discuss the latest advances in methods to Learn Representations of Graph Data. We start by identifying commonly used types of graph data and review basics of graph theory. This is followed by a discussion of the relationships between graph kernel methods and neural networks. Next we identify the major approaches used for learning representations of graph data namely: Kernel approaches, Convolutional approaches, Graph neural networks approaches, Graph embedding approaches and Probabilistic approaches. A variety of methods under each of the approaches are discussed and the survey is concluded with a brief discussion of the future of learning representation of graph data. Keywords Graph structured data, Graph representations, Graph-based neural networks, Graph embedding, Graph Convolutions.

A novel resampling framework is proposed to evaluate the robustness and generalization capability of deep learning models with respect to distribution shift. We use Auto Encoder Variational Bayes to find a latent representation of the data, on which a Variational Gaussian Mixture Model is applied to deliberately create distribution shift by dividing the dataset into different clusters. Wasserstein distance is used to characterize the extent of distribution shift between the training and the testing data splits. We compare several conventional Convolutional Neural Network (CNN) architectures as well as Bayesian CNN models for image classification on the Fashion-MNIST dataset to assess their robustness under the deliberately created distribution shift.

We introduce a pretraining technique called Selfie, which stands for SELF-supervised Image Embedding. Selfie generalizes the concept of masked language modeling to continuous data, such as images. Given masked-out patches in an input image, our method learns to select the correct patch, among other "distractor" patches sampled from the same image, to fill in the masked location. This classification objective sidesteps the need for predicting exact pixel values of the target patches. The pretraining architecture includes a network of convolutional blocks to process patches followed by an attention pooling network to summarize the content of unmasked patches before predicting masked ones. During finetuning, we reuse the convolutional weights found by pretraining. We evaluate our method on three benchmarks (CIFAR-10, ImageNet 32 x 32, and ImageNet 224 x 224) with varying amounts of labeled data, from 5% to 100% of the training sets. Our pretraining method provides consistent improvements to ResNet-50 across all settings compared to the standard supervised training of the same network. Notably, on ImageNet 224 x 224 with 60 examples per class (5%), our method improves the mean accuracy of ResNet-50 from 35.6% to 46.7%, an improvement of 11.1 points in absolute accuracy. Our pretraining method also improves ResNet-50 training stability, especially on low data regime, by significantly lowering the standard deviation of test accuracies across datasets.

Normalization methods play an important role in enhancing the performance of deep learning while their theoretical understandings have been limited. To theoretically elucidate the effectiveness of normalization, we quantify the geometry of the parameter space determined by the Fisher information matrix (FIM), which also corresponds to the local shape of the loss landscape under certain conditions. We analyze deep neural networks with random initialization, which is known to suffer from a pathologically sharp shape of the landscape when the network becomes sufficiently wide. We reveal that batch normalization in the last layer contributes to drastically decreasing such pathological sharpness if the width and sample number satisfy a specific condition. In contrast, it is hard for batch normalization in the middle hidden layers to alleviate pathological sharpness in many settings. We also found that layer normalization cannot alleviate pathological sharpness either. Thus, we can conclude that batch normalization in the last layer significantly contributes to decreasing the sharpness induced by the FIM.

We propose AutoGrow to automate depth discovery in Deep Neural Networks (DNNs): starting from a shallow seed architecture, AutoGrow grows new layers if the growth improves the accuracy; otherwise, the growth stops and the network depth is discovered. The residual and plain blocks are used as growing sub-modules to study DNNs with and without shortcuts. We propose generic growing and stopping policies to minimize human efforts spent on the optimal depth search. Our experiments show that by applying the same policy to different tasks, AutoGrow can always discover network depth effectively and achieve state-of-the-art accuracy on various datasets of MNIST, FashionMNIST, SVHN, CIFAR10, CIFAR100 and ImageNet. Comparing to Neural Architecture Search (NAS) that often designs a gigantic search space and consumes tremendous resources, AutoGrow lies at the other end of the research spectrum: it focuses on efficient depth discovery and reduces the growing and searching time to a level similar to that of training a single DNN. Thus, AutoGrow is able to scale up to large datasets such as ImageNet. Our study also reveals that previous Network Morphism is sub-optimal for increasing layer depth. Finally, we demonstrate that AutoGrow enables the training of deeper plain networks, which has been problematic even using Batch Normalization.

Recurrent neural networks (RNNs) are commonly applied to clinical time-series data with the goal of learning patient risk stratification models. Their effectiveness is due, in part, to their use of parameter sharing over time (i.e., cells are repeated hence the name recurrent). We hypothesize, however, that this trait also contributes to the increased difficulty such models have with learning relationships that change over time. Conditional shift, i.e., changes in the relationship between the input X and the output y, arises if the risk factors for the event of interest change over the course of a patient admission. While in theory, RNNs and gated RNNs (e.g., LSTMs) in particular should be capable of learning time-varying relationships, when training data are limited, such models often fail to accurately capture these dynamics. We illustrate the advantages and disadvantages of complete weight sharing (RNNs) by comparing an LSTM with shared parameters to a sequential architecture with time-varying parameters on three clinically-relevant prediction tasks: acute respiratory failure (ARF), shock, and in-hospital mortality. In experiments using synthetic data, we demonstrate how weight sharing in LSTMs leads to worse performance in the presence of conditional shift. To improve upon the dichotomy between complete weight sharing vs. no weight sharing, we propose a novel RNN formulation based on a mixture model in which we relax weight sharing over time. The proposed method outperforms standard LSTMs and other state-of-the-art baselines across all tasks. In settings with limited data, relaxed weight sharing can lead to improved patient risk stratification performance.

Neural Networks (NNs) have been found vulnerable to a class of imperceptible attacks, called adversarial examples, which arbitrarily alter the output of the network. These attacks have called the validity of NNs into question, particularly on sensitive problems such as medical imaging or fraud detection. We further argue that the fields of explainable AI and Human-In-The-Loop (HITL) algorithms are impacted by adversarial attacks, as attacks result in perturbations outside of the salient regions highlighted by state-of-the-art techniques such as LIME or Grad-CAM. This work accomplishes three things which greatly reduce the impact of adversarial examples, and pave the way for future HITL workflows: we propose a novel regularization technique inspired by the Lipschitz constraint which greatly improves an NN's resistance to adversarial examples; we propose a collection of novel network and training changes to complement the proposed regularization technique, including a Half-Huber activation function and an integrator-based controller for regularization strength; and we demonstrate that networks trained with this technique may be deliberately attacked to generate rich explanations. Our techniques led to networks more robust than the previous state of the art: using the Accuracy-Robustness Area (ARA), our most robust ImageNet classification network scored 42.2% top-1 accuracy on unmodified images and demonstrated an attack ARA of 0.0053, an ARA 2.4x greater than the previous state-of-the-art at the same level of accuracy on clean data, achieved with a network one-third the size. A far-reaching benefit of this technique is its ability to intuitively demonstrate decision boundaries to a human observer, allowing for improved debugging of NN decisions, and providing a means for improving the underlying model.