We develop the setting of sequential prediction based on shifting experts and on a "smooth" version of the method of specialized experts. To aggregate experts predictions, we use the AdaHedge algorithm, which is a version of the Hedge algorithm with adaptive learning rate, and extend it by the meta-algorithm Fixed Share. Due to this, we combine the advantages of both algorithms: (1) we use the shifting regret which is a more optimal characteristic of the algorithm; (2) regret bounds are valid in the case of signed unbounded losses of the experts. Also, (3) we incorporate in this scheme a "smooth" version of the method of specialized experts which allows us to make more flexible and accurate predictions. All results are obtained in the adversarial setting -- no assumptions are made about the nature of data source. We present results of numerical experiments for short-term forecasting of electricity consumption based on a real data.

Deep Neural Networks have been shown to be beneficial for a variety of tasks, in particular allowing for end-to-end learning and reducing the requirement for manual design decisions. However, still many parameters have to be chosen in advance, also raising the need to optimize them. One important, but often ignored system parameter is the selection of a proper activation function. Thus, in this paper we target to demonstrate the importance of activation functions in general and show that for different tasks different activation functions might be meaningful. To avoid the manual design or selection of activation functions, we build on the idea of genetic algorithms to learn the best activation function for a given task. In addition, we introduce two new activation functions, ELiSH and HardELiSH, which can easily be incorporated in our framework. In this way, we demonstrate for three different image classification benchmarks that different activation functions are learned, also showing improved results compared to typically used baselines.

This paper presents a useful method to achieve classification in satellite imagery. The approach is based on pixel level study employing various features such as correlation, homogeneity, energy and contrast. In this study gray-scale images are used for training the classification model. For supervised classification, two classification techniques are employed namely the Support Vector Machine (SVM) and the Naive Bayes. With textural features used for gray-scale images, Naive Bayes performs better with an overall accuracy of 76% compared to 68% achieved by SVM. The computational time is evaluated while performing the experiment with two different window sizes i.e., 50x50 and 70x70. The required computational time on a single image is found to be 27 seconds for a window size of 70x70 and 45 seconds for a window size of 50x50.

Wi-Fi) and the second/third/fourth generation (2/3/4G) mobile network, the number of mobile phones, which is 7.74 billion, 103.5 per 100 inhabitants all over the world in 2017, is rising dramatically [1]. Nowadays, mobile phone can not only send voice and text messages, but also easily and conveniently access the Internet which has been recognized as the most revolutionary development of Mobile Internet (M-Internet). Meanwhile, worldwide active mobile-broadband subscriptions in 2017 have increased to 4.22 billion, which is 9.21% higher than that in 2016 [1]. Figure 1 shows the numbers of mobile-cellular telephone and active mobile-broadband subscriptions of the world and main districts from 2010 to 2017. The numbers which are up to the bars are the mobile-cellular telephone or active mobile-broadband subscriptions (million) in the world of the year which increase each year. Under the M-Internet, various kinds of content (image, voice, video, etc.) can be sent and received everywhere and the related applications emerge to satisfy people's requirements, including working, study, daily life, entertainment, education, healthcare, etc. In China, mobile applications giants, i.e., Baidu, Alibaba and Tencent, held 78% of M-Internet online time per day in App which was about 2,412 minutes in 2017 [2]. This figure indicates that M-Internet has entered a rapidly growth stage.

In this paper we study the problem of acoustic scene classification, i.e., categorization of audio sequences into mutually exclusive classes based on their spectral content. We describe the methods and results discovered during a competition organized in the context of a graduate machine learning course; both by the students and external participants. We identify the most suitable methods and study the impact of each by performing an ablation study of the mixture of approaches. We also compare the results with a neural network baseline, and show the improvement over that. Finally, we discuss the impact of using a competition as a part of a university course, and justify its importance in the curriculum based on student feedback.

There is much existing research in the fields of affective computing and psychology which aims to predict what people are thinking and feeling from their facial expressions and other physiological data [10,11,17]. Much of it is based on the highly problematic "Classical Theory of Emotions", in which emotions are believed to be essential, discrete reactions of our bodies to changes in our environment [4], and therefore should be detectable in individuals by a machine once it learns what that emotion looks like for a general population. The research of Ekman and Friesen [6] is often held up as proof that this is how emotions work, and that, at least for the "basic" emotions that Ekman and Friesen "identified" (happiness, anger, sadness, surprise, fear, and disgust), they should be recognizable by anyone. Thus, typically, affective computing relies on supervised learning systems using large databases of multi-subject data. The initially accurate results of this approach seem to support the Classical Theory of Emotions, but inter-database testing (testing a system trained on one database on a different database) can result in significant decreases in accuracy [8, 9, 12, 14] (particularly for "spontaneous" or "non-posed" subject data), suggesting that the assumptions of universality made by the Classical Theory of Emotion are wrong. Our work is based on the newer "Theory of Constructed Emotions," (TCE) [4] which is a more scientifically provable and consistent framework for understanding human emotion and affect. According to neuroscientist Dr. Lisa Feldman Barrett, originator and lead proponent of the theory, emotions are mental concepts we learn at a young age which help us predict the meaning of things in our environment, not automatic, pre-wired response systems in our brains that react the same way for all people. Barrett says the only way we can accurately predict what another person is feeling from looking at their face or other physiological signals is by also being given some contextual information: what is happening to that person when they make that face?

We present an approximate Bayesian inference approach for estimating the intensity of a inhomogeneous Poisson process, where the intensity function is modelled using a Gaussian process (GP) prior via a sigmoid link function. Augmenting the model using a latent marked Poisson process and P\'olya--Gamma random variables we obtain a representation of the likelihood which is conjugate to the GP prior. We approximate the posterior using a free--form mean field approximation together with the framework of sparse GPs. Furthermore, as alternative approximation we suggest a sparse Laplace approximation of the posterior, for which an efficient expectation--maximisation algorithm is derived to find the posterior's mode. Results of both algorithms compare well with exact inference obtained by a Markov Chain Monte Carlo sampler and standard variational Gauss approach, while being one order of magnitude faster.

In this paper, we provide an information-theoretic interpretation of the Vector Quantized-Variational Autoencoder(VQ-VAE). We show that the loss function of the original VQ-VAE [1] can be derived from the variational deterministic information bottleneck (VDIB) principle [2]. On the other hand, the VQ-VAE trained by the Expectation Maximization (EM) algorithm [3] can be viewed as an approximation to the variational information bottleneck(VIB) principle [4]. I Introduction The recent advances of variational autoencoder(VAE) provide new unsupervised approaches to learn hidden structure of the data [5]. The variational autoencoder is a powerful generative model which allows inference of the learned latent representation. However, the classic VAEs are prone to the "posterior collapse "phenomenon that the latent representations are ignored due to the powerful decoder. Vector quantized variational autoencoder (VQ-VAE) learns discrete representations by incorporating the idea of vector quantization into the bottleneck stage and the "posterior collapse "can be avoided [1].

Using artificial neural network for the prediction of heat demand has attracted more and more attention. Weather conditions, such as ambient temperature, wind speed and direct solar irradiance, have been identified as key input parameters. In order to further improve the model accuracy, it is of great importance to understand the influence of different parameters. Based on an Elman neural network (ENN), this paper investigates the impact of direct solar irradiance and wind speed on predicting the heat demand of a district heating network. Results show that including wind speed can generally result in a lower overall mean absolute percentage error (MAPE) (6.43%) than including direct solar irradiance (6.47%); while including direct solar irradiance can achieve a lower maximum absolute deviation (71.8%) than including wind speed (81.53%). In addition, even though including both wind speed and direct solar irradiance shows the best overall performance (MAPE=6.35%).

In this paper, we continue our previous work on the Dirichlet mixture model (DMM)-based VQ to derive the performance bound of the LSF VQ. The LSF parameters are transformed into the $\Delta$LSF domain and the underlying distribution of the $\Delta$LSF parameters are modelled by a DMM with finite number of mixture components. The quantization distortion, in terms of the mean squared error (MSE), is calculated with the high rate theory. The mapping relation between the perceptually motivated log spectral distortion (LSD) and the MSE is empirically approximated by a polynomial. With this mapping function, the minimum required bit rate for transparent coding of the LSF is estimated.