Courses The major educational initiative of the JHUDSL is to create open-source online courses delivered through a range of platforms including Youtube, Github, Leanpub, and Coursera. Welcome to the "Introduction to Deep Learning" course! In the first week you'll learn about linear models and stochatic optimization methods. Please note that this is an advanced course and we assume basic knowledge of machine learning. I am currently working as a data science researcher and trainee at Jheronimus Academy of Data Science.

Deep learning techniques have been shown to be extremely effective for various classification and regression problems, but quantifying the uncertainty of their predictions and separating them into the epistemic and aleatoric fractions is still considered challenging. In oil and gas exploration projects, tools consisting of seismic, sonic, magnetic resonance, resistivity, dielectric and/or nuclear sensors are sent downhole through boreholes to probe the earth's rock and fluid properties. The measurements from these tools are used to build reservoir models that are subsequently used for estimation and optimization of hydrocarbon production. Machine learning algorithms are often used to estimate the rock and fluid properties from the measured downhole data. Quantifying uncertainties of these properties is crucial for rock and fluid evaluation and subsequent reservoir optimization and production decisions. These machine learning algorithms are often trained on a "ground-truth" or core database. During the inference phase which involves application of these algorithms to field data, it is critical that the machine learning algorithm flag data as out of distribution from new geologies that the model was not trained upon. It is also highly important to be sensitive to heteroscedastic aleatoric noise in the feature space arising from the combination of tool and geological conditions. Understanding the source of the uncertainty and reducing them is key to designing intelligent tools and applications such as automated log interpretation answer products for exploration and field development. In this paper we describe a methodology consisting of a system of dual networks comprising of the combination of a Bayesian Neural Network (BNN) and an Artificial Neural Network (ANN) addressing this challenge for geophysical applications.

Fisheye cameras are commonly used in applications like autonomous driving and surveillance to provide a large field of view ($>180^\circ$). However, they come at the cost of strong non-linear distortion which require more complex algorithms. In this paper, we explore Euclidean distance estimation on fisheye cameras for automotive scenes. Obtaining accurate and dense depth supervision is difficult in practice, but self-supervised learning approaches show promising results and could potentially overcome the problem. We present a novel self-supervised scale-aware framework for learning Euclidean distance and ego-motion from raw monocular fisheye videos without applying rectification. While it is possible to perform piece-wise linear approximation of fisheye projection surface and apply standard rectilinear models, it has its own set of issues like re-sampling distortion and discontinuities in transition regions. To encourage further research in this area, we will release this dataset as part of our WoodScape project \cite{yogamani2019woodscape}. We further evaluated the proposed algorithm on the KITTI dataset and obtained state-of-the-art results comparable to other self-supervised monocular methods. Qualitative results on an unseen fisheye video demonstrate impressive performance, see https://youtu.be/Sgq1WzoOmXg .

Momentum is a simple and popular technique in deep learning for gradient-based optimizers. We propose a decaying momentum (Demon) rule, motivated by decaying the total contribution of a gradient to all future updates. Applying Demon to Adam leads to significantly improved training, notably competitive to momentum SGD with learning rate decay, even in settings in which adaptive methods are typically non-competitive. Similarly, applying Demon to momentum SGD rivals momentum SGD with learning rate decay, and in many cases leads to improved performance. Demon is trivial to implement and incurs limited extra computational overhead, compared to the vanilla counterparts.

Precise estimation of uncertainty in predictions for AI systems is a critical factor in ensuring trust and safety. Conventional neural networks tend to be overconfident as they do not account for uncertainty during training. In contrast to Bayesian neural networks that learn approximate distributions on weights to infer prediction confidence, we propose a novel method, Information Robust Dirichlet networks, that learns the Dirichlet distribution on prediction probabilities by minimizing the expected $L_p$ norm of the prediction error and an information divergence loss that penalizes information flow towards incorrect classes, while simultaneously maximizing differential entropy of small adversarial perturbations to provide accurate uncertainty estimates. Properties of the new cost function are derived to indicate how improved uncertainty estimation is achieved. Experiments using real datasets show that our technique outperforms state-of-the-art neural networks, by a large margin, for estimating in-distribution and out-of-distribution uncertainty, and detecting adversarial examples.

Textual network embedding aims to learn low-dimensional representations of text-annotated nodes in a graph. Prior works have typically focused on fixed graph structures. However, real-world networks are often dynamic. We address this challenge with a novel end-to-end node-embedding model, called Dynamic Embedding for Textual Networks with a Gaussian Process (DetGP). Because the structure is allowed to be dynamic, our method uses the Gaussian process to take advantage of its non-parametric properties. After training, DetGP can be applied efficiently to dynamic graphs without re-training or backpropagation. To use both local and global graph structures, diffusion is used to model multiple hops between neighbors. The relative importance of global versus local structure for the embeddings is learned automatically. With the non-parametric nature of the Gaussian process, updating the embeddings for a changed graph structure requires only a forward pass through the learned model. Experiments demonstrate the empirical effectiveness of our method compared to baseline approaches, on link prediction and node classification. We further show that DetGP can be straightforwardly and efficiently applied to dynamic textual networks.

Neural backdoor attack is emerging as a severe security threat to deep learning, while the capability of existing defense methods is limited, especially for complex backdoor triggers. In the work, we explore the space formed by the pixel values of all possible backdoor triggers. An original trigger used by an attacker to build the backdoored model represents only a point in the space. It then will be generalized into a distribution of valid triggers, all of which can influence the backdoored model. Thus, previous methods that model only one point of the trigger distribution is not sufficient. Getting the entire trigger distribution, e.g., via generative modeling, is a key to effective defense. However, existing generative modeling techniques for image generation are not applicable to the backdoor scenario as the trigger distribution is completely unknown. In this work, we propose max-entropy staircase approximator (MESA), an algorithm for high-dimensional sampling-free generative modeling and use it to recover the trigger distribution. We also develop a defense technique to remove the triggers from the backdoored model. Our experiments on Cifar10 dataset demonstrate the effectiveness of MESA in modeling the trigger distribution and the robustness of the proposed defense method.

Large language models have recently achieved state of the art performance across a wide variety of natural language tasks. Meanwhile, the size of these models and their latency have significantly increased, which makes their usage costly, and raises an interesting question: do language models need to be large? We study this question through the lens of model compression. We present a novel, structured pruning approach based on low rank factorization and augmented Lagrangian l 0 norm regularization. Our structured approach achieves significant inference speedups while matching or outperforming our unstructured pruning baseline at various sparsity levels. We apply our method to state of the art models on the enwiki8 dataset and obtain a 1.19 perplexity score with just 5M parameters, vastly outperforming a model of the same size trained from scratch. We also demonstrate that our method can be applied to language model fine-tuning by pruning the BERT model on several downstream classification benchmarks. 1 1 Introduction Recent advances in language modeling have led to remarkable improvements on a variety of natural language tasks. These models, however, have grown increasingly large (Dai et al., 2019), rendering them slow and costly.

Recent studies on catastrophic forgetting during sequential learning typically focus on fixing the accuracy of the predictions for a previously learned task. In this paper we argue that the outputs of neural networks are subject to rapid changes when learning a new data distribution, and networks that appear to "forget" everything still contain useful representation towards previous tasks. Instead of enforcing the output accuracy to stay the same, we propose to reduce the effect of catastrophic forgetting on the representation level, as the output layer can be quickly recovered later with a small number of examples. Towards this goal, we propose an experimental setup that measures the amount of representational forgetting, and develop a novel meta-learning algorithm to overcome this issue. The proposed meta-learner produces weight updates of a sequential learning network, mimicking a multi-task teacher network's representation. We show that our meta-learner can improve its learned representations on new tasks, while maintaining a good representation for old tasks.

Given a state-of-the-art deep neural network text classifier, we show the existence of a universal and very small perturbation vector (in the embedding space) that causes natural text to be misclassified with high probability. Unlike images on which a single fixed-size adversarial perturbation can be found, text is of variable length, so we define the "universality" as "token-agnostic", where a single perturbation is applied to each token, resulting in different perturbations of flexible sizes at the sequence level. W e propose an algorithm to compute universal adversarial perturbations, and show that the state-of-the-art deep neural networks are highly vulnerable to them, even though they keep the neighborhood of tokens mostly preserved. W e also show how to use these adversarial perturbations to generate adversarial text samples. The surprising existence of universal "token-agnostic" adversarial perturbations may reveal important properties of a text classifier.