Defending models against Natural Language Processing adversarial attacks is a challenge because of the discrete nature of the text dataset. However, given the variety of Natural Language Processing applications, it is important to make text processing models more robust and secure. This paper aims to develop techniques that will help text processing models such as BERT to combat adversarial samples that contain misspellings. These developed models are more robust than off the shelf spelling checkers.

An image processing method for automatic quality assurance of Ericsson products is developed. The method consists of taking an image of the product, extract the product labels from the image, OCR the product numbers and make a database lookup to match the mounted product with the customer specification. The engineering innovation of the method developed in this report is that the OCR is performed using machine learning techniques. It is shown that machine learning can produce results that are on par or better than baseline OCR methods. The advantage with a machine learning based approach is that the associated neural network can be trained for the specific input images from the Ericsson factory. Imperfections in the image quality and varying type fonts etc. can be handled by properly training the net, a task that would have been very difficult with legacy OCR algorithms where poor OCR results typically need to be mitigated by improving the input image quality rather than changing the algorithm.

<p>The rapid expansion of intra-vehicle networks has increased the number of threats to such networks. Most modern vehicles implement various physical and data-link layer technologies. Vehicles are becoming increasingly autonomous and connected. Controller Area Network (CAN) is a serial bus system that is used to connect sensors and controllers (Electronic Control Units – ECUs) within a vehicle. ECUs vary widely in processing power, storage, memory, and connectivity. The goal of this research is to design, implement, and test an efficient and effective intrusion detection system for intra-vehicle CANs. Such a system must be capable of detecting intrusions in almost real-time with minimal resources. The research proposes a specific type of recursive neural network called Long Short-Term Memory (LSTM) to detect anomalies. It also proposes a decision engine that will use LSTM-classified anomalies to detect intrusions by using multiple contextual parameters. We have conducted multiple experiments on the optimal choice of various LSTM hyperparameters. We have tested our classification algorithm and our decision engine using data from real automobiles. We will present the results of our experiments and analyze our findings. After detailed evaluation of our intrusion detection system, we believe that we have designed a vehicle security solution that meets all the outlined requirements and goals.</p>

As part of Volvo Penta's initiative to further the development of predictive maintenance in their field test environments, this thesis compares neural networks in an effort to predict the occurrence of three common diagnostics trouble codes using field test data. To quantify the neural networks' performances for comparison a number of evaluation metrics were used. By training a multitude of differently configured feedforward neural networks with the processed field test data and evaluating the resulting models, it was found that the resulting models perform better than that of a baseline classifier. As such it is possible to use Volvo Penta's field test data along with neural networks to achieve predictive maintenance. It was also found that Long Short-Term Memory (LSTM) networks with methodically selected hyperparameters were able to predict the diagnostic trouble codes with the greatest performance among all the tested neural networks.

The interest towrds machine learning applied to turbulence has experienced a fast-paced growth in the last years. Thanks to deep-learning algorithms, flow-control stratigies have been designed, as well as tools to model and reproduce the most relevant turbulent features. In particular, the success of recurrent neural networks (RNNs) has been demonstrated in many recent studies and applications. The main objective of this project is to assess the capability of these networks to reproduce the temporal evolution of a minimal turbulent channel flow. We first obtain a data-driven model based on a modal decomposition in the Fourier domain (FFT-POD) on the time series sampled from the flow. This particular case of turbulent flow allows us to accurately simulate the most relevant coherent structures close to the wall. Long-short-term-memory (LSTM) networks and a Koopman-based framework (KNF) are trained to predict the temporal dynamics of the minimal channel flow modes. Tests with different configurations highlight the limits of the KNF method compared to the LSTM, given the complexity of the data-driven model. Long-term prediction for LSTM show excellent agreement from the statistical point of view, with errors below 2% for the best models. Furthermore, the analysis of the chaotic behaviour thorugh the use of the Lyapunov exponent and of the dynamic behaviour through Pointcaré maps emphasizes the ability of LSTM to reproduce the nature of turbulence. Alternative reduced-order models (ROMS), based on the identification of different turbulent structures, are explored and they continue to show a good potential in predicting the temporal dynamics of the minimal channel.

Different road users follow different behaviors and intentions in the trajectories that they traverse. Predicting the intent of these road users at intersections would not only help increase the comfort of drive in autonomous vehicles, but also help detect potential accidents. In this thesis, the research objective is to build models that predicts future positions of road users (pedestrians,cyclists and autonomous shuttles) by capturing behaviors endemic to different road users.  Firstly, a constant velocity state space model is used as a benchmark for intent prediction, with a fresh approach to estimate parameters from the data through the EM algorithm. Then, a neural network based LSTM sequence modeling architecture is used to better capture the dynamics of road user movement and their dependence on the spatial area. Inspired by the recent success of transformers and attention in text mining, we then propose a mechanism to capture the road users' social behavior amongst their neighbors. To achieve this, past trajectories of different road users are forward propagated through the LSTM network to obtain representative feature vectors for each road users' behaviour. These feature vectors are then passed through an attention-layer to obtain representations that incorporate information from other road users' feature vectors, which are in-turn used to predict future positions for every road user in the frame. It is seen that the attention based LSTM model slightly outperforms the plain LSTM models, while both substantially outperform the constant velocity model. A comparative qualitative analysis is performed to assess the behaviors that are captured/missed by the different models. The thesis concludes with a dissection of the behaviors captured by the attention module.