Abstract Semantic understanding of roadways is a key enabling factor for safe autonomous driving. However, existing autonomous driving datasets provide well-structured urban roads while ignoring unstructured roadways containing distress, potholes, water puddles, and various kinds of road patches i.e., earthen, gravel etc. To this end, we introduce Road Region Segmentation dataset (R2S100K)—a large-scale dataset and benchmark for training and evaluation of road segmentation in aforementioned challenging unstructured roadways. R2S100K comprises 100K images extracted from a large and diverse set of video sequences covering more than 1000 km of roadways. Out of these 100K privacy respecting images, 14,000 images have fine pixel-labeling of road regions, with 86,000 unlabeled images that can be leveraged through semi-supervised learning methods. Alongside, we present an Efficient Data Sampling based self-training framework to improve learning by leveraging unlabeled data. Our experimental results demonstrate that the proposed method significantly improves learning methods in generalizability and reduces the labeling cost for semantic segmentation tasks. Our benchmark will be publicly available to facilitate future research at https://r2s100k.github.io/ .

To combat the rapid spread of misinformation on social networks, automated misinformation detection systems based on deep neural networks (DNNs) have been developed. However, these tools are often proprietary and lack transparency, which limits their usefulness. Furthermore, privacy concerns limit data sharing by data owners as well as by data-driven misinformation-detection services. Although data encryption techniques can help address privacy concerns in DNN inference, there is a challenge to the seamless integration of these techniques due to the encryption errors induced by cascaded encrypted operations, as well as a mismatch between the tools used for DNNs and cryptography. In this paper, we make two-fold contributions. First, we study the noise bounds of homomorphic encryption (HE) operations as error propagation in DNN layers and derive two properties that, if satisfied by the layer, will considerably reduce the output error. We identify that$L_{2}$regularization and sigmoid activation satisfy these properties and validate our hypothesis, for instance, replacing ReLU with sigmoid reduced the output error by$10^{6}\times$(best case) to$10\times$(worst case). Second, we extend the Python encryption library TenSeal by enabling the automatic conversion of a TensorFlow DNN into an encryption-compatible DNN with a few lines of code. These contributions are significant as encryption-friendly DL architectures are sorely needed to close the gap between DL-in-research and DL-in-practice.

The metaverse is a nascent concept that envisions a virtual universe, a collaborative space where individuals can interact, create, and participate in a wide range of activities. Privacy in the metaverse is a critical concern as the concept evolves and immersive virtual experiences become more prevalent. The metaverse privacy problem refers to the challenges and concerns surrounding the privacy of personal information and data within Virtual Reality (VR) environments as the concept of a shared VR space becomes more accessible. Metaverse will harness advancements from various technologies such as Artificial Intelligence (AI), Extended Reality (XR) and Mixed Reality (MR) to provide personalized and immersive services to its users. Moreover, to enable more personalized experiences, the metaverse relies on the collection of fine-grained user data that leads to various privacy issues. Therefore, before the potential of the metaverse can be fully realized, privacy concerns related to personal information and data within VR environments must be addressed. This includes safeguarding users’ control over their data, ensuring the security of their personal information, and protecting in-world actions and interactions from unauthorized sharing. In this paper, we explore various privacy challenges that future metaverses are expected to face, given their reliance on AI for tracking users, creating XR and MR experiences, and facilitating interactions. Moreover, we thoroughly analyze technical solutions such as differential privacy, Homomorphic Encryption, and Federated Learning and discuss related sociotechnical issues regarding privacy.

With the advent of modern smart grid networks, advanced metering infrastructure provides real-time information from smart meters (SM) and sensors to energy companies and consumers. The smart grid is indeed a paradigm that is enabled by the Internet of Things (IoT) and in which the SM acts as an IoT device that collects and transmits data over the Internet to enable intelligent applications. However, IoT data communicated over the smart grid could however be maliciously altered, resulting in energy theft due to unbilled energy consumption. Machine learning (ML) techniques for energy theft detection (ETD) based on IoT data are promising but are nonetheless constrained by the poor quality of data and particularly its imbalanced nature (which emerges from the dominant representation of honest users and poor representation of the rare theft cases). Leading ML-based ETD methods employ synthetic data generation to balance the training the dataset. However, these are trained to maximise average correct detection instead of ETD. In this work, we formulate an energy-aware evaluation framework that guides the model training to maximise ETD and minimise the revenue loss due to mis-classification. We propose a convolution neural network with positive bias (CNN-B) and another with focal loss CNN (CNN-FL) to mitigate the data imbalance impact. These outperform the state of the art and the CNN-B achieves the highest ETD and the minimum revenue loss with a loss reduction of 30.4% compared to the highest loss incurred by these methods.

The use of artificial intelligence (AI) at the edge is transforming every aspect of the lives of human beings from scheduling daily activities to personalized shopping recommendations. Since the success of AI is to be measured ultimately in terms of how it benefits human beings, and that the data driving the deep learning-based edge AI algorithms are intricately and intimately tied to humans, it is important to look at these AI technologies through a human-centric lens. However, despite the significant impact of AI design on human interests, the security and trustworthiness of edge AI applications are not foolproof and ethicalneither foolproof nor ethical; Moreover, social norms are often ignored duringin the design, implementation, and deployment of edge AI systems. In this paper, we make the following two contributions: Firstly, we analyze the application of edge AI through a human-centric perspective. More specifically, we present a pipeline to develop human-centric embedded machine learning (HC-EML) applications leveraging a generic human-centric AI (HCAI) framework. Alongside, we also analyzediscuss the privacy, trustworthiness, robustness, and security aspects of HC-EML applications with an insider look at their challenges and possible solutions along the way. Secondly, to illustrate the gravity of these issues, we present a case study on the task of human facial emotion recognition (FER) based on AffectNet dataset, where we analyze the effects of widely used input quantization on the security, robustness, fairness, and trustworthiness of an EML model. We find that input quantization partially degrades the efficacy of adversarial and backdoor attacks at the cost of a slight decrease in accuracy over clean inputs. By analyzing the explanations generated by SHAP, we identify that the decision of a FER model is largely influenced by features such as eyes, alar crease, lips, and jaws. Additionally, we note that input quantization is notably biased against the dark skin faces, and hypothesize that low-contrast features of dark skin faces may be responsible for the observed trends. We conclude with precautionary remarks and guidelines for future researchers.

In recent years, advancements in machine learning (ML) techniques, in particular, deep learning (DL) methods have gained a lot of momentum in solving inverse imaging problems, often surpassing the performance provided by hand-crafted approaches. Traditionally, analytical methods have been used to solve inverse imaging problems such as image restoration, inpainting, and superresolution. Unlike analytical methods for which the problem is explicitly defined and the domain knowledge is carefully engineered into the solution, DL models do not benefit from such prior knowledge and instead make use of large datasets to predict an unknown solution to the inverse problem. Recently, a new paradigm of training deep models using a single image, named untrained neural network prior (UNNP) has been proposed to solve a variety of inverse tasks, e.g., restoration and inpainting. Since then, many researchers have proposed various applications and variants of UNNP. In this paper, we present a comprehensive review of such studies and various UNNP applications for different tasks and highlight various open research problems which require further research.

Deep machine learning techniques have shown promising results in network traffic classification, however, the robustness of these techniques under adversarial threats is still in question. Deep machine learning models are found vulnerable to small carefully crafted adversarial perturbations posing a major question on the performance of deep machine learning techniques. In this paper, we propose a black-box adversarial attack on network traffic classification. The proposed attack successfully evades deep machine learning-based classifiers which highlights the potential security threat of using deep machine learning techniques to realize autonomous networks.