USCIS data show the "Buy American and Hire American" executive order signed by Donald Trump on April 18, 2017, set in motion policies that made it more difficult for foreign-born scientists and engineers to gain H-1B visas and other types of employer-sponsored visas. At the same time the Trump administration has launched a trade war premised, in part, on worry that China will pass the U.S. in the technologies of the future, Trump officials are guaranteeing America will not have the talent to produce such technologies by blocking the entry of foreign-born scientists and engineers. New data show that in 2017, U.S. Citizenship and Immigration Services (USCIS) denied many more H-1B petitions and issued numerous costly Requests for Evidence to stop highly educated foreign nationals from working in America. "Soon after Donald Trump issued the'Buy American and Hire American' presidential executive order on April 18, 2017, U.S. Citizenship and Immigration Services started to increase both the Requests for Evidence (RFEs) and denials for H-1B petitions for high-skilled foreign-born professionals," according to a new report from the National Foundation for American Policy (NFAP). Data obtained by NFAP from U.S. Citizenship and Immigration Services show the impact of the administration's efforts, particularly in the 4th quarter of Fiscal Year 2017 (which started July 1, 2017).

China's momentum and challenges in artificial intelligence investments yield telling lessons for its worldwide observers. China's ambition in artificial intelligence is often framed as a tech rivalry between two important centers for digital innovation -- the east coast of China and the U.S. West Coast. But this rivalry is an undercard for the main event: AI's largest and most enduring contributions will be in non-technology sectors, as traditional companies unlock value in regions far away from Silicon Valley and the string of coastal cities that constitute China's innovation corridor. For non-tech sectors, our research indicates that Chinese companies' approach to adopting AI differs from those in other regions, raising important questions. Buoyed by the country's latest five-year plan and enabled by centralized data, these companies are investing aggressively in AI and adapting their business models to accommodate for AI's potential.

JSFoo is India's premier JavaScript conference. HasGeek creates discussion spaces for geeks.

For different language pairs, word-level neural machine translation (NMT) models with a fixed-size vocabulary suffer from the same problem of representing out-of-vocabulary (OOV) words. The common practice usually replaces all these rare or unknown words with a 〈UNK〉 token, which limits the translation performance to some extent. Most of recent work handled such a problem by splitting words into characters or other specially extracted subword units to enable open-vocabulary translation. Byte pair encoding (BPE) is one of the successful attempts that has been shown extremely competitive by providing effective subword segmentation for NMT systems. In this paper, we extend the BPE style segmentation to a general unsupervised framework with three statistical measures: frequency (FRQ), accessor variety (AV) and description length gain (DLG). We test our approach on two translation tasks: German to English and Chinese to English. The experimental results show that AV and DLG enhanced systems outperform the FRQ baseline in the frequency weighted schemes at different significant levels.

Given two networks with the same training loss on a dataset, when would they have drastically different test losses and errors? Better understanding of this question of generalization may improve practical applications of deep networks. In this paper we show that with cross-entropy loss it is surprisingly simple to induce significantly different generalization performances for two networks that have the same architecture, the same meta parameters and the same training error: one can either pretrain the networks with different levels of "corrupted" data or simply initialize the networks with weights of different Gaussian standard deviations. A corollary of recent theoretical results on overfitting shows that these effects are due to an intrinsic problem of measuring test performance with a cross-entropy/exponential-type loss, which can be decomposed into two components both minimized by SGD -- one of which is not related to expected classification performance. However, if we factor out this component of the loss, a linear relationship emerges between training and test losses. Under this transformation, classical generalization bounds are surprisingly tight: the empirical/training loss is very close to the expected/test loss. Furthermore, the empirical relation between classification error and normalized cross-entropy loss seem to be approximately monotonic

Matrix factorization is a key tool in data analysis; its applications include recommender systems, correlation analysis, signal processing, among others. Binary matrices are a particular case which has received significant attention for over thirty years, especially within the field of data mining. Dictionary learning refers to a family of methods for learning overcomplete basis (also called frames) in order to efficiently encode samples of a given type; this area, now also about twenty years old, was mostly developed within the signal processing field. In this work we propose two binary matrix factorization methods based on a binary adaptation of the dictionary learning paradigm to binary matrices. The proposed algorithms focus on speed and scalability; they work with binary factors combined with bit-wise operations and a few auxiliary integer ones. Furthermore, the methods are readily applicable to online binary matrix factorization. Another important issue in matrix factorization is the choice of rank for the factors; we address this model selection problem with an efficient method based on the Minimum Description Length principle. Our preliminary results show that the proposed methods are effective at producing interpretable factorizations of various data types of different nature.

Neural networks are very powerful learning systems, but they do not readily generalize from one task to the other. This is partly due to the fact that they do not learn in a compositional way, that is, by discovering skills that are shared by different tasks, and recombining them to solve new problems. In this paper, we explore the compositional generalization capabilities of recurrent neural networks (RNNs). We first propose the lookup table composition domain as a simple setup to test compositional behaviour and show that it is theoretically possible for a standard RNN to learn to behave compositionally in this domain when trained with standard gradient descent and provided with additional supervision. We then remove this additional supervision and perform a search over a large number of model initializations to investigate the proportion of RNNs that can still converge to a compositional solution. We discover that a small but non-negligible proportion of RNNs do reach partial compositional solutions even without special architectural constraints. This suggests that a combination of gradient descent and evolutionary strategies directly favouring the minority models that developed more compositional approaches might suffice to lead standard RNNs towards compositional solutions.

We study the problem of synthesizing a number of likely future frames from a single input image. In contrast to traditional methods that have tackled this problem in a deterministic or non-parametric way, we propose to model future frames in a probabilistic manner. Our probabilistic model makes it possible for us to sample and synthesize many possible future frames from a single input image. To synthesize realistic movement of objects, we propose a novel network structure, namely a Cross Convolutional Network; this network encodes image and motion information as feature maps and convolutional kernels, respectively. In experiments, our model performs well on synthetic data, such as 2D shapes and animated game sprites, and on real-world video frames. We present analyses of the learned network representations, showing it is implicitly learning a compact encoding of object appearance and motion. We also demonstrate a few of its applications, including visual analogy-making and video extrapolation.

The presence of gender stereotypes in many aspects of society is a well-known phenomenon. In this paper, we focus on studying and quantifying such stereotypes and bias in the Man Bookers Prize winning fiction. We consider 275 books shortlisted for Man Bookers Prize between 1969 and 2017. The gender bias is analyzed by semantic modeling of book descriptions on Goodreads. This reveals the pervasiveness of gender bias and stereotype in the books on different features like occupation, introductions and actions associated to the characters in the book.

The goal of point set registration is to find point-by-point correspondences between point sets, each of which characterizes the shape of an object. Because local preservation of object geometry is assumed, prevalent algorithms in the area can often elegantly solve the problems without using geometric information specific to the objects. This means that registration performance can be further improved by using prior knowledge of object geometry. In this paper, we propose a novel point set registration method using the Gaussian mixture model with prior shape information encoded as a statistical shape model. Our transformation model is defined as a combination of the similarity transformation, motion coherence, and the statistical shape model. Therefore, the proposed method works effectively if the target point set includes outliers and missing regions, or if it is rotated. The computational cost can be reduced to linear, and therefore the method is scalable to large point sets. The effectiveness of the method will be verified through comparisons with existing algorithms using datasets concerning human body shapes, hands, and faces.