The capability to detect objects is a core part of autonomous driving. Due to sensor noise and incomplete data, perfectly detecting and localizing every object is infeasible. Therefore, it is important for a detector to provide the amount of uncertainty in each prediction. Providing the autonomous system with reliable uncertainties enables the vehicle to react differently based on the level of uncertainty. Previous work has estimated the uncertainty in a detection by predicting a probability distribution over object bounding boxes. In this work, we propose a method to improve the ability to learn the probability distribution by considering the potential noise in the ground-truth labeled data. Our proposed approach improves not only the accuracy of the learned distribution but also the object detection performance.

Thermography is a passive and non-contact imaging technique used extensively in the medical arena, but in relation to breast care, it has not been accepted as being on a par with mammography. This paper proposes the analysis of thermograms with the use of artificial neural networks (ANN) and bio-statistical methods, including regression and receiver operating characteristics (ROC). It is desired that through these approaches, highly accurate diagnosis using thermography techniques can be achieved. The suggested method is a multi-pronged approach comprising of linear regression, radial basis function network (RBFN) and ROC analysis. It is a novel, integrative and powerful technique that can be used to analyse large amounts of complicated measured data such as temperature values extracted from abnormal and healthy breast thermograms. The use of regression allows the correlation between the variables and the actual health status of the subject, which is decided by other traditional means such as the gold standard of mammography for breast cancer detection. This is important as it helps to select the appropriate variables to be used as inputs for building the neural network. RBFN is next trained to produce the desired outcome that is either positive or negative. When this is done, the RBFN possess the ability to predict the outcome when there are new input variables. The advantages of using RBFN include fast training of superior classification and decision-making abilities as compared to other networks such as backpropagation. Lastly, ROC is applied to evaluate the sensitivity, specificity and accuracy of the outcome for the RBFN test files. The proposed technique has an accuracy rate of 80.95%, with 100% sensitivity and 70.6% specificity in identifying breast cancer. The results are promising as compared to clinical examination by experienced radiologists, which has an accuracy rate of approximately 60-70%. To sum up, technological advances in the field of infrared thermography over the last 20 years warrant a re-evaluation of the use of high-resolution digital thermographic camera systems in the diagnosis and management of breast cancer. Thermography seeks to identify the presence of a tumour by the elevated temperature associated with increase blood flow and cellular activity. Of particular interest would be investigation in younger women and men, for whom mammography is either unsuitable or of limited effectiveness. The paper evaluated the high-definition digital infrared thermographic technology and knowledge base; and supports the development of future diagnostic and therapeutic services in breast cancer imaging. Through the use of integrative ANN and bio-statistical methods, advances are made in thermography application with regard to achieving a higher level of consistency. For breast cancer care, it has become possible to use thermography as a powerful adjunct and biomarker tool, together with mammography for diagnosis purposes.

According to the efficient coding hypothesis, sensory systems are adapted to maximize their ability to encode information about the environment. Sensory neurons play a key role in encoding by selectively modulating their firing rate for a subset of all possible stimuli. This pattern of modulation is often summarized via a tuning curve. The optimally efficient distribution of tuning curves has been calculated in variety of ways for one-dimensional (1-D) stimuli. However, many sensory neurons encode multiple stimulus dimensions simultaneously. It remains unclear how applicable existing models of 1-D tuning curves are for neurons tuned across multiple dimensions. We describe a mathematical generalization that builds on prior work in 1-D to predict optimally efficient multidimensional tuning curves. Our results have implications for interpreting observed properties of neuronal populations. For example, our results suggest that not all tuning curve attributes (such as gain and bandwidth) are equally useful for evaluating the encoding efficiency of a population.

Detecting the intention of drivers is an essential task in self-driving, necessary to anticipate sudden events like lane changes and stops. Turn signals and emergency flashers communicate such intentions, providing seconds of potentially critical reaction time. In this paper, we propose to detect these signals in video sequences by using a deep neural network that reasons about both spatial and temporal information. Our experiments on more than a million frames show high per-frame accuracy in very challenging scenarios.

Thermography is a non-invasive and non-contact imaging technique widely used in the medical arena. This paper investigates the analysis of thermograms with the use of biostatistical methods and Artificial Neural Networks (ANN). It is desired that through these approaches, highly accurate diagnosis using thermography techniques can be established. The proposed advanced technique is a multipronged approach comprising of Linear Regression (LR), Radial Basis Function Network (RBFN) and Receiver Operating Characteristics (ROC). It is a novel and integrative technique that can be used to analyze complicated and large numerical data. In this study, the advanced technique will be used to analyze breast cancer thermogram for diagnosis purposes. The use of LR shows the correlation between the variables and the actual health status (healthy or cancerous) of the subject, which is decided by using mammography. This is important when selecting the variables to be used as inputs, in particular, for building the neural network. For ANN, RBFN is applied. Based on the various inputs fed into the network, RBFN will be trained to produce the desired outcome, which is either positive for cancerous or negative for healthy cases. When this is done, the RBFN algorithm will possess the ability to predict the outcome when there are new input variables. The advantages of using RBFN include fast training, superior classification and decision making abilities as compared to other networks such as back-propagation. Next, ROC is used to evaluate the accuracy, sensitivity and specificity of the outcome of RBFN Test files. The best results obtained are an accuracy (score) rate of 80.95%, with 81.2% sensitivity and 88.2% specificity. For breast cancer diagnosis, clinical examination by experienced doctors has an accuracy rate of approximately 60–70%. Hence, the proposed method has a higher accuracy rate than the existing practice. Through the use of Bio-statistical methods and ANN, improvements are made in thermography application with regard to achieving a higher level of accuracy rate in diagnosis as compared to clinical examination. It has now become possible to use thermography as a powerful adjunct tool for breast cancer detection, together with mammography for diagnosis purposes.