Datasets in engineering applications are often limited and contaminated, mainly due to unavoidable measurement noise and signal distortion. Thus, using conventional data-driven approaches to build a reliable discriminative model, and further applying this identified surrogate to uncertainty analysis remains to be very challenging. A deep learning approach is presented to provide predictions based on limited and noisy data. To address noise perturbation, the Bayesian learning method that naturally facilitates an automatic updating mechanism is considered to quantify and propagate model uncertainties into predictive quantities. Specifically, hierarchical Bayesian modeling (HBM) is first adopted to describe model uncertainties, which allows the prior assumption to be less subjective, while also makes the proposed surrogate more robust. Next, the Bayesian inference is seamlessly integrated into the DL framework, which in turn supports probabilistic programming by yielding a probability distribution of the quantities of interest rather than their point estimates. Variational inference (VI) is implemented for the posterior distribution analysis where the intractable marginalization of the likelihood function over parameter space is framed in an optimization format, and stochastic gradient descent method is applied to solve this optimization problem. Finally, Monte Carlo simulation is used to obtain an unbiased estimator in the predictive phase of Bayesian inference, where the proposed Bayesian deep learning (BDL) scheme is able to offer confidence bounds for the output estimation by analyzing propagated uncertainties. The effectiveness of Bayesian shrinkage is demonstrated in improving predictive performance using contaminated data, and various examples are provided to illustrate concepts, methodologies, and algorithms of this proposed BDL modeling technique.

Time series with non-uniform intervals occur in many applications, and are difficult to model using standard recurrent neural networks (RNNs). We generalize RNNs to have continuous-time hidden dynamics defined by ordinary differential equations (ODEs), a model we call ODE-RNNs. Furthermore, we use ODE-RNNs to replace the recognition network of the recently-proposed Latent ODE model. Both ODE-RNNs and Latent ODEs can naturally handle arbitrary time gaps between observations, and can explicitly model the probability of observation times using Poisson processes. We show experimentally that these ODE-based models outperform their RNN-based counterparts on irregularly-sampled data.

Robotic vision is a field where continual learning can play a significant role. An embodied agent operating in a complex environment subject to frequent and unpredictable changes is required to learn and adapt continuously. In the context of object recognition, for example, a robot should be able to learn (without forgetting) objects of never seen classes as well as improving its recognition capabilities as new instances of already known classes are discovered. Ideally, continual learning should be triggered by the availability of short videos of single objects and performed online on onboard hardware. In this paper, we introduce a novel fine-grained continual learning protocol based on the CORe50 benchmark and propose two continual learning techniques that can learn effectively even in the challenging case of nearly 400 small non-i.i.d.

Sequence-to-sequence (seq2seq) based ASR systems have shown state-of-the-art performances while having clear advantages in terms of simplicity. However, comparisons are mostly done on speaker independent (SI) ASR systems, though speaker adapted conventional systems are commonly used in practice for improving robustness to speaker and environment variations. In this paper, we apply speaker adaptation to seq2seq models with the goal of matching the performance of conventional ASR adaptation. Specifically, we investigate Kullback-Leibler divergence (KLD) as well as Linear Hidden Network (LHN) based adaptation for seq2seq ASR, using different amounts (up to 20 hours) of adaptation data per speaker. Our SI models are trained on large amounts of dictation data and achieve state-of-the-art results. We obtained 25% relative word error rate (WER) improvement with KLD adaptation of the seq2seq model vs. 18.7% gain from acoustic model adaptation in the conventional system. We also show that the WER of the seq2seq model decreases log-linearly with the amount of adaptation data. Finally, we analyze adaptation based on the minimum WER criterion and adapting the language model (LM) for score fusion with the speaker adapted seq2seq model, which result in further improvements of the seq2seq system performance.

In order to keep track of the operational state of power grid, the world's largest sensor systems, smart grid, was built by deploying hundreds of millions of smart meters. Such system makes it possible to discover and make quick response to any hidden threat to the entire power grid. Non-technical losses (NTLs) have always been a major concern for its consequent security risks as well as immeasurable revenue loss. However, various causes of NTL may have different characteristics reflected in the data. Accurately capturing these anomalies faced with such large scale of collected data records is rather tricky as a result. In this paper, we proposed a new methodology of detecting abnormal electricity consumptions. We did a transformation of the collected time-series data which turns it into an image representation that could well reflect users' relatively long term consumption behaviors. Inspired by the excellent neural network architecture used for objective detection in computer vision domain, we designed our deep learning model that takes the transformed images as input and yields joint featured inferred from the multiple aspects the input provides. Considering the limited labeled samples, especially the abnormal ones, we used our model in a semi-supervised fashion that is brought out in recent years. The model is tested on samples which are verified by on-field inspections and our method showed significant improvement.

Recently, a residual network (ResNet) with a single residual block has been shown to outperform linear predictors, in the sense that all its local minima are at least as good as the best linear predictor. We take a step towards extending this result to deep ResNets. As motivation, we first show that there exist datasets for which all local minima of a fully-connected ReLU network are no better than the best linear predictor, while a ResNet can have strictly better local minima. Second, we show that even at its global minimum, the representation obtained from the residual blocks of a 2-block ResNet does not necessarily improve monotonically as more blocks are added, highlighting a fundamental difficulty in analyzing deep ResNets. Our main result on deep ResNets shows that (under some geometric conditions) any critical point is either (i) at least as good as the best linear predictor; or (ii) the Hessian at this critical point has a strictly negative eigenvalue. Finally, we complement our results by analyzing near-identity regions of deep ResNets, obtaining size-independent upper bounds for the risk attained at critical points as well as the Rademacher complexity.

For instance, players with a very rapid in-game Nowadays most video games are played online and every progression (who reach a high level after a relatively short action by every player is recorded. This generates extremely playtime, regardless of their lifetime) and low spend might rich datasets that--with the aid of machine learning be overlooked by traditional segmentation methods due to techniques--can provide deep insights on user behavior, their lack of direct economic value; however, these are the including accurate predictions of the future actions of each most skillful players, and a careful study of their traits and player. Increasingly diverse demographics are now playing behavior--allowed by our approach--could provide developers games in a highly competitive market. Furthermore, we are with a lot of useful insights.

The use of computational methods to evaluate aesthetics in photography has gained interest in recent years due to the popularization of convolutional neural networks and the availability of new annotated datasets. Most studies in this area have focused on designing models that do not take into account individual preferences for the prediction of the aesthetic value of pictures. We propose a model based on residual learning that is capable of learning subjective, user specific preferences over aesthetics in photography, while surpassing the state-of-the-art methods and keeping a limited number of user-specific parameters in the model. Our model can also be used for picture enhancement, and it is suitable for content-based or hybrid recommender systems in which the amount of computational resources is limited.

Auto dealerships receive thousands of calls daily from customers interested in sales, service, vendors and jobseekers. With so many calls, it is very important for auto dealers tounderst and the intent of these calls to provide positive customer experiences that ensure customer satisfaction, deeper customer engagement to boost sales and revenue, and optimum allocation of agents or customer service representatives across the business. In this paper, we define the problem of customer phone call intent as a multi-class classification problem stemming from the large database of recorded phone call transcripts. To solve this problem, we develop a convolutional neural network (CNN)-based supervised learning model to classify the customer calls into four intent categories: sales, service, vendor or jobseeker. Experimental results show that with the thrust of our scalable data labeling method to provide sufficient training data, the CNN-based predictive model performs very well on long text classification according to tests that measure the model's quantitative metrics of F1-Score, precision, recall, and accuracy.

Ball trajectory data are one of the most fundamental and useful information in the evaluation of players' performance and analysis of game strategies. Although vision-based object tracking techniques have been developed to analyze sport competition videos, it is still challenging to recognize and position a high-speed and tiny ball accurately. In this paper, we develop a deep learning network, called TrackNet, to track the tennis ball from broadcast videos in which the ball images are small, blurry, and sometimes with afterimage tracks or even invisible. The proposed heatmap-based deep learning network is trained to not only recognize the ball image from a single frame but also learn flying patterns from consecutive frames. TrackNet takes images with a size of $640\times360$ to generate a detection heatmap from either a single frame or several consecutive frames to position the ball and can achieve high precision even on public domain videos. The network is evaluated on the video of the men's singles final at the 2017 Summer Universiade, which is available on YouTube. The precision, recall, and F1-measure of TrackNet reach $99.7\%$, $97.3\%$, and $98.5\%$, respectively. To prevent overfitting, 9 additional videos are partially labeled together with a subset from the previous dataset to implement 10-fold cross-validation, and the precision, recall, and F1-measure are $95.3\%$, $75.7\%$, and $84.3\%$, respectively. A conventional image processing algorithm is also implemented to compare with TrackNet. Our experiments indicate that TrackNet outperforms conventional method by a big margin and achieves exceptional ball tracking performance. The dataset and demo video are available at https://nol.cs.nctu.edu.tw/ndo3je6av9/.