Abstract
Deep learning has achieved significant improvements in various tasks in Computer Vision. However, acquiring a large number of the dataset is a challenge in real-world applications, especially if they are new class objects for Deep Learning. Furthermore, the distribution of classes in the dataset is often imbalanced - a bottleneck of the neural network’s performance in classification. One possible real-world application is road segmentation, which is crucial for autonomous driving and sophisticated driver assistance systems to comprehend the driving environment. Recent years have seen significant advancements in road segmentation with the help of Deep Learning. Inaccurate road boundaries and lighting fluctuations such as shadows and overexposed zones are still challenging issues. Prediction performance is also impacted by an improper class distribution, which arises because most image pixels belong to the background (negative class), while the goal is to identify road pixels (positive class). In this paper, we focus on the topic of "visual road classification," where the target is to label each pixel as containing either a road or a background. We tackle this task by implementing a novel Adaptive Augmentation algorithm by integrating with some recently suggested encoder-decoder based convolutional neural network architecture and compare the qualitative and quantitative experimental results with traditional augmentation algorithm. The proposed method uses an adaptive augmentation module to improve performance under improper class distribution conditions. Experimental results show that the suggested method achieves higher segmentation accuracy than state-of-the-art methods on the KITTI road detection benchmark datasets.
Index Terms: Road Segmentation; Autonomous Vehicle Design; Image Segmentation; Image Processing; RGB Processing
Abstract
In this work, we present an end-to-end framework with a predictor model that provides classifier outputs from given input features and an adversary that tries to predict protected or sensitive features in order to mitigate intrinsic biases with respect to sensitive features (e.g., race, sex). Our proposed model increases the predictor’s capacity to produce correct predictions, while decreasing the adversary’s ability to anticipate sensitive features. We include a novel Semantic Attention (SA) module to the framework and demonstrate that our SA based Generative Adversarial Network (SAGAN) is able to significantly minimize the adversary’s capability to predict sensitive features, while retaining the predictor’s predictive accuracy. UCI Adult (Census) dataset was used a benchmark dataset for the testing. Our results demonstrate that the predictive model does not lose much accuracy, while achieving a Disparate Impact (DI) score very close to 1. The flexibility of the method makes it fitting to be applicable to a broad spectrum of gradient-based learning models, including both regression and classification tasks as well as different definitions of fairness. The source code for the implementation is available on github [1].
Index Terms: SAGAN, Semantic Attention, GAN
Abstract
The Software-Defined Network (SDN) is a next-generation network that uses OpenFlow to decouple the control plane from the data plane of forwarding devices. Other protocols for southbound interfaces include ForCES and POF. However, some security issues might be in action on the SDN, so that attackers can take control of the SDN control plane. Since live video calling, QoS control, high bandwidth needs, and resource management are inevitable in any SDN/Software-Defined Cellular Network (SDCN), traffic monitoring is an integral approach for safeguarding against DDoS, heavy hitters, and superspreaders. In such a scenario, SDN traffic measurement comes into action. Thus, we survey SDN traffic measurement solutions to assess how these solutions can make a secure, efficient, and robust SDN/SDCN architecture. This research classifies SDN traffic measurement solutions according to network application behavior and compares several ML approaches. Furthermore, we find out the challenges related to SDN/SDCN traffic measurement and the future scope of research, which will guide the design and development of more advanced traffic measurement solutions for a scalable, heterogeneous, hierarchical, and widely deployed SDN/SDCN architecture. In more detail, we list different kinds of practical machine learning (ML) approaches to analyze how we can improve traffic measurement performances. We conclude that using ML in SDN traffic measurement solutions will help secure SDNs/SDCNs in complementary ways.
Keywords: SDN; SDN traffic measurement; SDN measurement; SDCN
Abstract
Coreference resolution is well-studied in NLP; however, Bengali coreference resolution research has not been as well investigated as it has been for English and other rich languages. Bengali has a richer morphology than English despite having fewer resources. This research introduce a tiny new dataset of coreference annotations over Bengali texts from four domains in this article called BenCo. In this dataset, there are 48,610 tokens and 5488 mention annotations structured into 470 mention clusters. The article outline the procedure used to generate this dataset and use it to build an end-to-end neural network-based system. This study will help to clarify how Bengali coreference phenomena differ depending on the domain and inspires others to produce more materials in the language. Also, insufficient information transfer in the zero-shot multilingual situations, which may indicate the need for language-specific resources for this job.
Index Terms: Computational Linguistics; Natural Language Processing; Coreference Resolution; Anaphora Resolution; Cataphora Resolution