The operation of large utility companies such as Consolidated Edison Company of New York, Inc. (Con Edison) typically rely on large quantities of regulation documents from external institutions which inform the company of upcoming or ongoing policy changes or new requirements the company might need to comply with if deemed applicable. As a concrete example, if a recent regulatory publication mentions that the timeframe for the Company to respond to a reported system emergency in its service territory changes from within X time to within Y time-then the affected operating groups will be notified, and internal Company operating procedures may need to be reviewed and updated accordingly to comply with the new regulatory requirement. Each such regulation document needs to be reviewed manually by an expert to determine if the document is relevant to the company and, if so, which department it is relevant to. In order to help enterprises improve the efficiency of their operation, we propose an automatic document classification pipeline that determines whether a document is important for the company or not, and if deemed important it forwards those documents to the departments within the company for further review. Binary classification task of determining the importance of a document is done via ensembling the Naive Bayes (NB), support vector machine (SVM), random forest (RF), and artificial neural network (ANN) together for the final prediction, whereas the multi-label classification problem of identifying the relevant departments for a document is executed by the transformer-based DocBERT model. We apply our pipeline to a large corpus of tens of thousands of text data provided by Con Edison and achieve an accuracy score over 80%. Compared with existing solutions for document classification which rely on a single classifier, our paper i) ensemble multiple classifiers for better accuracy results and escaping from the problem of overfitting, ii) utilize pretrained transformer-based DocBERT model to achieve ideal performance for multi-label classification task and iii) introduce a bi-level structure to improve the performance of the whole pipeline where the binary classification module works as a rough filter before finally distributing the text to corresponding departments through the multi-label classification module.

Among various sensors for assisted and autonomous driving systems, automotive radar has been considered as a robust and low-cost solution even in adverse weather or lighting conditions. With the recent development of radar technologies and open-sourced annotated data sets, semantic segmentation with radar signals has become very promising. However, existing methods are either computationally expensive or discard significant amounts of valuable information from raw 3D radar signals by reducing them to 2D planes via averaging. In this work, we introduce ERASE-Net, an Efficient RAdar SEgmentation Network to segment the raw radar signals semantically. The core of our approach is the novel detectthen-segment method for raw radar signals. It first detects the center point of each object, then extracts a compact radar signal representation, and finally performs semantic segmentation. We show that our method can achieve superior performance on radar semantic segmentation task compared to the state-ofthe-art (SOTA) technique. Furthermore, our approach requires up to 20x less computational resources. Finally, we show that the proposed ERASE-Net can be compressed by 40% without significant loss in performance, significantly more than the SOTA network, which makes it a more promising candidate for practical automotive applications.

In this paper, we study the sharpness of a deep learning (DL) loss landscape around local minima in order to reveal systematic mechanisms underlying the generalization abilities of DL models. Our analysis is performed across varying network and optimizer hyper-parameters, and involves a rich family of different sharpness measures. We compare these measures and show that the low-pass filter-based measure exhibits the highest correlation with the generalization abilities of DL models, has high robustness to both data and label noise, and furthermore can track the double descent behavior for neural networks. We next derive the optimization algorithm, relying on the low-pass filter (LPF), that actively searches the flat regions in the DL optimization landscape using SGD-like procedure. The update of the proposed algorithm, that we call LPF-SGD, is determined by the gradient of the convolution of the filter kernel with the loss function and can be efficiently computed using MC sampling. We empirically show that our algorithm achieves superior generalization performance compared to the common DL training strategies. On the theoretical front, we prove that LPF-SGD converges to a better optimal point with smaller generalization error than SGD.

Securing enterprise networks presents challenges in terms of both their size and distributed structure. Data required to detect and characterize malicious activities may be diffused and may be located across network and endpoint devices. Further, cyber-relevant data routinely exceeds total available storage, bandwidth, and analysis capability, often by several orders of magnitude. Real-time detection of threats within or across very large enterprise networks is not simply an issue of scale, but also a challenge due to the variable nature of malicious activities and their presentations. The system seeks to develop a hierarchy of cyber reasoning layers to detect malicious behavior, characterize novel attack vectors and present an analyst with a contextualized human-readable output from a series of machine learning models. We developed machine learning algorithms for scalable throughput and improved recall for our Multi-Resolution Joint Optimization for Enterprise Security and Forensics (ESAFE) solution. This Paper will provide an overview of ESAFE's Machine Learning Modules, Attack Ontologies, and Automated Smart Alert generation which provide multi-layer reasoning over cross-correlated sensors for analyst consumption.

This paper focuses on understanding how the generalization error scales with the amount of the training data for deep neural networks (DNNs). Existing techniques in statistical learning theory require a computation of capacity measures, such as VC dimension, to provably bound this error. It is however unclear how to extend these measures to DNNs and therefore the existing analyses are applicable to simple neural networks, which are not used in practice, e.g., linear or shallow (at most two-layer) ones or otherwise multi-layer perceptrons. Moreover many theoretical error bounds are not empirically verifiable. In this paper we derive estimates of the generalization error that hold for deep networks and do not rely on unattainable capacity measures. The enabling technique in our approach hinges on two major assumptions: i) the network achieves zero training error, ii) the probability of making an error on a test point is proportional to the distance between this point and its nearest training point in the feature space and at certain maximal distance (that we call radius) it saturates. Based on these assumptions we estimate the generalization error of DNNs. The obtained estimate scales as $\mathcal{O}\left( {\frac{1}{{\delta {N^{1/d}}}}} \right)$, where N is the size of the training data, and is parameterized by two quantities, the effective dimensionality of the data as perceived by the network (d) and the aforementioned radius (δ), both of which we find empirically. We show that our estimates match with the experimentally-obtained behavior of the error on multiple learning tasks using benchmark data-sets and realistic models. Estimating training data requirements is essential for deployment of safety critical applications such as autonomous driving, medical diagnostics etc. Furthermore, collecting and annotating training data requires a huge amount of financial, computational and human resources. Our empirical estimates will help to efficiently allocate resources.

Melanoma is one of ten most common cancers in the US. Early detection is crucial for survival, but often the cancer is diagnosed in the fatal stage. Deep learning has the potential to improve cancer detection rates, but its applicability to melanoma detection is compromised by the limitations of the available skin lesion data bases, which are small, heavily imbalanced, and contain images with occlusions. We build deep-learning-based tools for data purification and augmentation to counter-act these limitations. The developed tools can be utilized in a deep learning system for lesion classification and we show how to build such system. The system heavily relies on the processing unit for removing image occlusions and the data generation unit, based on generative adversarial networks, for populating scarce lesion classes, or equivalently creating virtual patients with pre-defined types of lesions. We empirically verify our approach and show that incorporating these two units into melanoma detection system results in the superior performance over common baselines.

In many machine learning applications, from medical diagnostics to autonomous driving, the availability of prior knowledge can be used to improve the predictive performance of learning algorithms and incorporate `physical,' `domain knowledge,' or `common sense' concepts into training of machine learning systems as well as verify constraints/properties of the systems. We explore the learning using privileged information paradigm and show how to incorporate the privileged information, such as segmentation mask available along with the classification label of each example, into the training stage of convolutional neural networks. This is done by augmenting the CNN model with an architectural component that effectively focuses model's attention on the desired region of the input image during the training process and that is transparent to the network's label prediction mechanism at testing. This component effectively corresponds to the visualization strategy for identifying the parts of the input, often referred to as visualization mask, that most contribute to the prediction, yet uses this strategy in reverse to the classical setting in order to enforce the desired visualization mask instead. We verify our proposed algorithms through exhaustive experiments on benchmark ImageNet and PASCAL VOC data sets and achieve improvements in the performance of $2.4\%$ and $2.7\%$ over standard single-supervision model training. Finally, we confirm the effectiveness of our approach on skin lesion classification problem.