Population migration is valuable information which leads to proper decision in urban-planning strategy, massive investment, and many other fields. For instance, inter-city migration is a posterior evidence to see if the government's constrain of population works, and inter-community immigration might be a prior evidence of real estate price hike. With timely data, it is also impossible to compare which city is more favorable for the people, suppose the cities release different new regulations, we could also compare the customers of different real estate development groups, where they come from, where they probably will go. Unfortunately these data was not available. In this paper, leveraging the data generated by positioning team in Didi, we propose a novel approach that timely monitoring population migration from community scale to provincial scale. Migration can be detected as soon as in a week. It could be faster, the setting of a week is for statistical purpose. A monitoring system is developed, then applied nation wide in China, some observations derived from the system will be presented in this paper. This new method of migration perception is origin from the insight that nowadays people mostly moving with their personal Access Point (AP), also known as WiFi hotspot. Assume that the ratio of AP moving to the migration of population is constant, analysis of comparative population migration would be feasible. More exact quantitative research would also be done with few sample research and model regression. The procedures of processing data includes many steps: eliminating the impact of pseudo-migration AP, for instance pocket WiFi, and second-hand traded router; distinguishing moving of population with moving of companies; identifying shifting of AP by the finger print clusters, etc..

Methods from convex optimization are widely used as building blocks for deep learning algorithms. However, the reasons for their empirical success are unclear, since modern convolutional networks (convnets), incorporating rectifier units and max-pooling, are neither smooth nor convex. Standard guarantees therefore do not apply. This paper provides the first convergence rates for gradient descent on rectifier convnets. The proof utilizes the particular structure of rectifier networks which consists in binary active/inactive gates applied on top of an underlying linear network. The approach generalizes to max-pooling, dropout and maxout. In other words, to precisely the neural networks that perform best empirically. The key step is to introduce gated games, an extension of convex games with similar convergence properties that capture the gating function of rectifiers. The main result is that rectifier convnets converge to a critical point at a rate controlled by the gated-regret of the units in the network. Corollaries of the main result include: (i) a game-theoretic description of the representations learned by a neural network; (ii) a logarithmic-regret algorithm for training neural nets; and (iii) a formal setting for analyzing conditional computation in neural nets that can be applied to recently developed models of attention.

As another example, to achieve high accuracy in recommender systems [45], [60], we need to fully understand the content of items (e.g., documents and movies), analyze the profile and preference of users, and evaluate the similarity among users. Deep learning is good at the first subtask while PGM excels at the other two. Besides the fact that better understanding of item content would help with the analysis of user profiles, the estimated similarity among users could provide valuable information for understanding item content in return. In order to fully utilize this bidirectional effect to boost recommendation accuracy, we might wish to unify deep learning and PGM in one single principled probabilistic framework, as done in [60]. Besides recommender systems, the need for Bayesian deep learning may also arise when we are dealing with control of nonlinear dynamical systems with raw images as input. Consider controlling a complex dynamical system according to the live video stream received from a camera. This problem can be transformed into iteratively performing two tasks, perception from raw images and control based on dynamic models. The perception task can be taken care of using multiple layers of simple nonlinear transformation (deep learning) while the control task usually needs more sophisticated models like hidden Markov models and Kalman filters [21], [38]. The feedback loop is then completed by the fact that actions chosen by the control model can affect the received video stream in return.

We consider the problem of exact recovery of any $m\times n$ matrix of rank $\varrho$ from a small number of observed entries via the standard nuclear norm minimization framework. Such low-rank matrices have degrees of freedom $(m+n)\varrho - \varrho^2$. We show that any arbitrary low-rank matrices can be recovered exactly from a $\Theta\left(((m+n)\varrho - \varrho^2)\log^2(m+n)\right)$ randomly sampled entries, thus matching the lower bound on the required number of entries (in terms of degrees of freedom), with an additional factor of $O(\log^2(m+n))$. To achieve this bound on sample size we observe each entry with probabilities proportional to the sum of corresponding row and column leverage scores, minus their product. We show that this relaxation in sampling probabilities (as opposed to sum of leverage scores in Chen et al, 2014) can give us an $O(\varrho^2\log^2(m+n))$ additive improvement on the (best known) sample size obtained by Chen et al, 2014, for the nuclear norm minimization. Experiments on real data corroborate the theoretical improvement on sample size. Further, exact recovery of $(a)$ incoherent matrices (with restricted leverage scores), and $(b)$ matrices with only one of the row or column spaces to be incoherent, can be performed using our relaxed leverage score sampling, via nuclear norm minimization, without knowing the leverage scores a priori. In such settings also we can achieve improvement on sample size.

This work is motivated by the needs of predictive analytics on healthcare data as represented by Electronic Medical Records. Such data is invariably problematic: noisy, with missing entries, with imbalance in classes of interests, leading to serious bias in predictive modeling. Since standard data mining methods often produce poor performance measures, we argue for development of specialized techniques of data-preprocessing and classification. In this paper, we propose a new method to simultaneously classify large datasets and reduce the effects of missing values. It is based on a multilevel framework of the cost-sensitive SVM and the expected maximization imputation method for missing values, which relies on iterated regression analyses. We compare classification results of multilevel SVM-based algorithms on public benchmark datasets with imbalanced classes and missing values as well as real data in health applications, and show that our multilevel SVM-based method produces fast, and more accurate and robust classification results.

This article introduces a method for teaching the CAR and CDR extensions. In programming, a container has multiple cells for holding other objects. We use an access language to specify a cell and say what action to take when it is found, such as returning the object in the cell. As an example, in FORTRAN, the primary container is the array, and there is a simple sub-language inside of FORTRAN used for accessing an element within an array. Namely, symbol (index) Here, the parenthesis are literal punctuation, the symbol will represent the base address for an array, and the index will be multiplied by the element size and added to the base address for finding an element.

One of the most important tasks in Machine Learning are the Classification tasks (a.k.a. Classification is used to make an accurate prediction of the class of entries in the test set (a dataset of which the entries have not been labelled yet) with the model which was constructed from a training set. You could think of classifying crime in the field of Pre-Policing, classifying patients in the Health sector, classifying houses in the Real-Estate sector. Another field in which classification is big, is Natural Lanuage Processing (NLP). This is the field of science with the goal to makes machines (computers) understand (written) human language.

Bob, CFO of ABC Inc is about to get on an earnings call after just reporting a 20% miss on earnings due to slower revenue growth than forecasted. Company ABC's stock price is plummeting, down 25% in extended hour trading. The board is furious and investors demand answers on the discrepancies. Inaccurate revenue forecast remains one of the biggest risks for CFOs. In a recent study, more than 50% of companies feel their pipeline forecast is only about 50% accurate.

Emotional intelligence is a cornerstone of human interactions--an essential part of what it means to be human. But now, artificial intelligences are being developed to better read and process human emotions, which is already changing the way we interact with robots. In the early 1990s, psychologists Salovey and Mayer were the first to recognize emotional intelligence as a set of knowledge and skills distinct from other forms of intelligence, defining it as "the ability to monitor one's own and other's feelings and emotions, to discriminate among them, and to use this information to guide one's thinking and actions." Emotional intelligence is something that seems wonderfully and innately human. But it turns out the tenets of emotional intelligence--which we start picking up in infancy and which seem so closely linked to human nature itself--can be quantified and reduced to logical procedures and algorithms.

This series of tutorial is about evolutionary computation: what it is, how it works and how to implement it in your projects and games. At the end of this series you'll be able to harness the power of evolution to find the solution to problems you have no idea how to solve. As a toy example, this tutorial will show how evolutionary computation can be used to teach a simple creature to walk. If you want to try the power of evolutionary computation directly in your browser, try Genetic Algorithm Walkers. As a programmer, you might be familiar with the concept of algorithm.