Estimating an object’s position can be of great interest in several applications,and there exists many different methods to do so. One approach is with Directionof Arrival (DOA) measurements from receivers to use the triangulation techniqueto estimate one or more transmitter’s position. One algorithm which can find theDOA measurements from several transmitters is the MUltiple SIgnal Classification(MUSIC) algorithm. However, this still leaves a ambiguity problem which givesfalse solutions, so called ghost points, if the number of receivers is not sufficient.In this report solving this problem with the help of deep learning is studied. Thethesis’s main objective is to investigate and study whether it is possible to performpositioning with measurements from the MUSIC-algorithm using deep learningand image processing methods. A deep neural network is built in TensorFlow and trained and tested using datagenerated from MATLAB. This thesis’s setup consists of two receivers, which areused to locate two transmitters. The network uses two MUSIC spectra from thetwo receivers, and returns a probability distribution of where the transmittersare located. The results are compared with a traditional method and are analysed.The results presented in this thesis show that it is possible to perform positioningusing deep learning methods. However, there is a lot of room for improvementwith accuracy, which can be an important future research direction to explore.

500 of 690 training images used in optimized training runs.

Intelligent transportation systems (ITS) have become a popular method to enhance the safety and efficiency of transportation. Pedestrians, as an essential participant of ITS, are very vulnerable in a traffic collision, compared with the passengers inside the vehicle. In order to protect the safety of all traffic participants and enhance transportation efficiency, the novel autonomous vehicles are required to detect pedestrians accurately and timely. In the area of pedestrian detection, deep learning-based pedestrian detection methods have gained significant development since the appearance of powerful GPUs. A large number of researchers are paying efforts to improve the accuracy of pedestrian detection by utilizing the Convolutional Neural Network (CNN)-based detectors. In this thesis, we propose a one-stage anchor-free pedestrian detector named Bi-Center Network (BCNet), which is aided by the semantic features of pedestrians' visible parts. The framework of our BCNet has two main modules: the feature extraction module produces the concatenated feature maps that extracted from different layers of ResNet, and the four parallel branches in the detection module produce the full body center keypoint heatmap, visible part center keypoint heatmap, heights, and offsets, respectively. The final bounding boxes are converted from the high response points on the fused center keypoint heatmap and corresponding predicted heights and offsets. The fused center keypoint heatmap contains the semantic feature fusion of the full body and the visible part of each pedestrian. Thus, we conduct ablation studies and discover the efficiency of feature fusion and how visibility features benefit the detector's performance by proposing two types of approaches: introducing two weighting hyper-parameters and applying three different attention mechanisms. Our BCNet gains 9.82% MR-2 (the lower the better) on the Reasonable setup of the CityPersons dataset, compared to baseline model which gains 12.14% MR-2 . The experimental results indicate that the performance of pedestrian detection could be significantly improved because the visibility semantic could prompt stronger responses on the heatmap. We compare our BCNet with state-of-the-art models on the CityPersons dataset and ETH dataset, which shows that our detector is effective and achieves a promising performance.

Connected vehicle networks and future autonomous driving systems call for characterization of risky driving events to improve safety applications and autonomous driving features. Precision of driving event characterization (\gls{dec}) systems in connected vehicles has become increasingly important with the responsive connectivity that is available to the modern vehicles. While risky behavior patterns entail potential safety issues on road networks, the advent of vehicular sensing and vehicular networks cannot guarantee accurate characterization of driving/movement behavior of vehicles and the precision of such systems still remains an open issue. Additionally, artificial intelligence-backed solutions are vital components towards highly accurate characterization systems in the modern transportation. However, such solutions require significant volume of driving event data for an acceptable level of performance. With this in mind, the proposal of this thesis is three-fold: 1) a reliable methodology to generate representative data under the scarcity of diverse anomalous sensory data, 2) classification of mobility/driving events of vehicles with attention-based deep learning methods, and 3) a modular prior-knowledge input method to the characterization methodologies in order to further improve the trustworthiness of the systems. Implementing the proposed steps, we are able to not only increase the consistency in the training process but also reduce the false positive detection instances compared to the previous models. One of the roadblocks against robust event characterization systems in connected vehicles that is tackled in this thesis is the scarcity of anomalous driving data to make the training of event classification models robust. To mitigate this issue an optimized deep recurrent neural network-based encoding model is introduced to extract the precise feature representation of the anomalous data. The utilization of the encoded input to the previous network allowed for a 12\% accuracy improvement. Furthermore, we introduced a framework for precise risky driving behavior detection that takes advantage of an attention-based neural networks model. Ultimately, the combination of prior knowledge modelling with our network and some optimizations to the network structure, the model outperforms the state-of-the-art solutions by reaching an average accuracy of 0.96 and F1-score of 0.92.

Every year, drivers are spending over 3 billions to repair damage on vehicle caused by potholes. Along with the financial disaster, potholes cause frustration in drivers. Also, with the emerging development of automated vehicles, road safety with automation in mind is being a necessity. Deep Learning techniques offer intelligent alternatives to reduce the loss caused by spotting pothole. The world is connected in such a way that the information can be shared in no time. Using the power of connectivity, we can communicate the information of potholes to other vehicles and also the department of Transportation for necessary action. A significant number of research efforts have been done with a view to help detect potholes in the pavements. In this thesis, we have compared two object detection algorithms belonging to two major classes i.e. single shot detectors and two stage detectors using our dataset. Comparing the results in the Faster RCNN and YOLOv5, we concluded that, potholes take a small portion in image which makes potholes detection with YOLOv5 less accurate than the Faster RCNN, but keeping the speed of detection in mind, we have suggested that YOLOv5 will be a better solution for this task. Using the YOLOv5 model and image processing technique, we calculated approximate area of potholes and visualized the shape of potholes. Thus obtained information can be used by the Department of Transportation for planning necessary construction tasks. Also, we can use these information to warn the drivers about the severity of potholes depending upon the shape and area.

In this paper, we try to understand the concept of ’Neural Collapse’ from a mathemati-cal point of view. The survey will be conducted based on [1]. The authors of [1] providea first global optimization landscape analysis of Neural Collapse. Mainly there are threeaspects the authors like to investigate. The first is to add the weight decay on classicalcross-entropy loss to show that the global minimizers are the simplex ETF based onanalysing the Hessian. Secondly, the ’Layer-peeled’ network still preserves the im-portant features of the full network. In other words even simplifying the loss functionthe network does not lose its explainability. Lastly, how the Layer-peeled network canreduce the memory costs and generalization is as good as the full network. Our studydelves into these details on, how the simplified network is defined? How this simplifiednetwork is different from the original network in terms of the loss function, and finallywe understand the theory behind these steps. We also conduct numerical analysis onspecific input, observe and analyze this phenomenon and finally report our results.

Knowing metastasis is the primary cause of cancer-related deaths incentivized research to unravel the complex cellular processes that drive the metastasis. Advancement in technology and specifically the advent of high-throughput sequencing provides knowledge of such processes. This knowledge led to the development of therapeutic and clinical applications. In this regard, predicting metastasis onset has also been explored using artificial intelligence (AI) approaches that are machine learning (ML), and more recently, deep learning (DL). This thesis discusses the revolutionary field of ML/DL and its applications in cancer metastasis prediction. We are raising the question of whether there is a better way to improve the prediction of metastasis? We effectively addressed this by reviewing strides made in this regard in current literature to draw some conclusions based on a comprehensive review. Then, we used this knowledge to develop multiple ML/DL models using different omics data types that can accurately and cost-effectively predict if the cancer is in the metastatic state and suggest the metastasis site. Beyond that, we show the biological functions that the DL model uses to perform the prediction. We proved that ML/DL could improve efficiency and diagnostic accuracy and can be used to develop novel predictors of prognosis despite some existing challenges.

Despite the significant progress in 3D vision in recent years, collecting large amounts of high-quality 3D data remains a challenge. Hence, developing solutions to extract 3D object information efficiently is a significant problem. We aim for an effective shape classification algorithm to facilitate accurate recognition and efficient search of sizeable 3D model databases. This thesis has two contributions in this space: a) a novel meta-learning approach for 3D object recognition and b) propose a new compositional 3D recognition task and dataset. For 3D recognition, we proposed a few-shot semi-supervised meta-learning model based on Pointnet++ representation with a prototypical random walk loss. In particular, we developed the random walk semi-supervised loss that enables fast learning from a few labeled examples by enforcing global consistency over the data manifold and magnetizing unlabeled points around their class prototypes. On the compositional recognition front, we create a large-scale, richly annotated stylized dataset called 3D CoMPaT. This large dataset primarily focuses on stylizing 3D shapes at part-level with compatible materials. We introduce Grounded CoMPaT Recognition as the task of collectively recognizing and grounding compositions of materials on parts of 3D Objects.

No description available.

Background. The rapid growth of IoT devices in homes put people at risk of cyberattacks and the low power and computing capabilities in IoT devices make it difficultto design a security solution for them. One method of preventing cyber attacks isan Intrusion Detection System (IDS) that can identify incoming attacks so that anappropriate action can be taken. Previous attempts have been made using machinelearning and deep learning however these attempts have struggled at detecting newattacks.Objectives. In this work we use a convolutional neural network IoTNet designed forIoT devices to classify network attacks. In order to evaluate the use of deep learningin intrusion detection systems on IoT.Methods. The neural network was trained on the NF-UNSW-NB15-v2 datasetwhich contains 9 different types of attacks. We used a method that transformedthe network flow data into RGB images which were fed to the neural network forclassification. We compared IoTNet to a basic convolutional neural network as abaseline.Results. The results show that IoTNet did not perform better at classifying networkattacks when compared to a basic convolutional neural network. It also showed thatboth network had low precision for most classes.Conclusions. We found that IoTNet is unfit to be used as an intrusion detectionsystem in the general case and that further research must be done in order to improvethe precision of the neural network.