Recognizing pedestrian attributes is an important task in computer vision community due to it plays an important role in video surveillance. Many algorithms has been proposed to handle this task. The goal of this paper is to review existing works using traditional methods or based on deep learning networks. Firstly, we introduce the background of pedestrian attributes recognition (PAR, for short), including the fundamental concepts of pedestrian attributes and corresponding challenges. Secondly, we introduce existing benchmarks, including popular datasets and evaluation criterion. Thirdly, we analyse the concept of multi-task learning and multi-label learning, and also explain the relations between these two learning algorithms and pedestrian attribute recognition. We also review some popular network architectures which have widely applied in the deep learning community. Fourthly, we analyse popular solutions for this task, such as attributes group, part-based, \emph{etc}. Fifthly, we shown some applications which takes pedestrian attributes into consideration and achieve better performance. Finally, we summarized this paper and give several possible research directions for pedestrian attributes recognition. The project page of this paper can be found from the following website: \url{https://sites.google.com/view/ahu-pedestrianattributes/}.

The use of orthogonal projections on high-dimensional input and target data in learning frameworks is studied. First, we investigate the relations between two standard objectives in dimension reduction, maximizing variance and preservation of pairwise relative distances. The derivation of their asymptotic correlation and numerical experiments tell that a projection usually cannot satisfy both objectives. In a standard classification problem we determine projections on the input data that balance them and compare subsequent results. Next, we extend our application of orthogonal projections to deep learning frameworks. We introduce new variational loss functions that enable integration of additional information via transformations and projections of the target data. In two supervised learning problems, clinical image segmentation and music information classification, the application of the proposed loss functions increase the accuracy.

Recent studies on diffusion-based sampling methods have shown that Langevin Monte Carlo (LMC) algorithms can be beneficial for non-convex optimization, and rigorous theoretical guarantees have been proven for both asymptotic and finite-time regimes. Algorithmically, LMC-based algorithms resemble the well-known gradient descent (GD) algorithm, where the GD recursion is perturbed by an additive Gaussian noise whose variance has a particular form. Fractional Langevin Monte Carlo (FLMC) is a recently proposed extension of LMC, where the Gaussian noise is replaced by a heavy-tailed {\alpha}-stable noise. As opposed to its Gaussian counterpart, these heavy-tailed perturbations can incur large jumps and it has been empirically demonstrated that the choice of {\alpha}-stable noise can provide several advantages in modern machine learning problems, both in optimization and sampling contexts. However, as opposed to LMC, only asymptotic convergence properties of FLMC have been yet established. In this study, we analyze the non-asymptotic behavior of FLMC for non-convex optimization and prove finite-time bounds for its expected suboptimality. Our results show that the weak-error of FLMC increases faster than LMC, which suggests using smaller step-sizes in FLMC. We finally extend our results to the case where the exact gradients are replaced by stochastic gradients and show that similar results hold in this setting as well.

An important part of many machine learning workflows on graphs is vertex representation learning, i.e., learning a low-dimensional vector representation for each vertex in the graph. Recently, several powerful techniques for unsupervised representation learning have been demonstrated to give the state-of-the-art performance in downstream tasks such as vertex classification and edge prediction. These techniques rely on random walks performed on the graph in order to capture its structural properties. These structural properties are then encoded in the vector representation space. However, most contemporary representation learning methods only apply to static graphs while real-world graphs are often dynamic and change over time. Static representation learning methods are not able to update the vector representations when the graph changes; therefore, they must re-generate the vector representations on an updated static snapshot of the graph regardless of the extent of the change in the graph. In this work, we propose computationally efficient algorithms for vertex representation learning that extend random walk based methods to dynamic graphs. The computation complexity of our algorithms depends upon the extent and rate of changes (the number of edges changed per update) and on the density of the graph. We empirically evaluate our algorithms on real world datasets for downstream machine learning tasks of multi-class and multi-label vertex classification. The results show that our algorithms can achieve competitive results to the state-of-the-art methods while being computationally efficient.

As artificial intelligence use cases expand inside all types of businesses and industries, you can expect another sequel to a timeless IT hiring story: The Skills Shortage, AI edition – coming soon to a team near you. As Pat Calhoun, CEO of Espressive, told us recently: "Most organizations want to embrace AI as part of their digital transformation but do not have the developers, AI experts, and linguists to develop their own or to even train the engines of pre-built solutions to deliver on the promise." As new technologies emerge and especially as their adoption begins to spike, there's commonly a gap between an organization's goals and the technical skills necessary to achieve those goals. So while AI might sound exotic and fancy, you don't actually need a newfangled playbook to bridge the gap. "Despite its evocative title, AI is remarkably similar to other fields in information technology, in that success comes through continuous learning, training, and great processes," says Zachary Jarvinen, head of technology strategy for AI and analytics at OpenText.

I remember having written a response that was specifically focused on Andrew Ng's Deep Learning training that was launched with a lot of fanfare in October last year. I have added excepts from my Quora answer here and there and this is me just visiting my own answers based on my year long experience since June 2017 working with CEOs and Chair(wo)men of large enterprises, training about 9000 people in my classical (meaning hands-on workshops where we learn the old fashioned way face-to-face) and interacting with tens of thousands of learners worldwide. I will however be brutally honest about my initial observation of the first 1.5 weeks -- which I went through yesterday with great anticipation and truly enjoyed (still enjoying!), of what I experienced. This may actually not have anything to do with his capabilities or intentions rather it("the dilemma") owes this to latest trend (pretty much close to madness) of packing a deep learning course in a MOOC and try to teach to folks everything in bunch of nutshells. I'll get to that in a minute, but first my quick analysis of who this Deep Learning course / specialization may or may not be for. So, who might this course be for?

We are announcing the release of our state-of-the-art off-policy model-free reinforcement learning algorithm, soft actor-critic (SAC). This algorithm has been developed jointly at UC Berkeley and Google Brain, and we have been using it internally for our robotics experiment. Soft actor-critic is, to our knowledge, one of the most efficient model-free algorithms available today, making it especially well-suited for real-world robotic learning. We also release our implementation of SAC, which is particularly designed for real-world robotic systems. What makes an ideal deep RL algorithm for real-world systems?

Artificial intelligence promises to have a dramatic--and yes, disruptive--effect on U.S. education and jobs in the next decade. But that technology won't be entirely homegrown: Chinese companies, particularly those building products or services laced with the machine learning algorithms, are increasingly playing a role in the tools that we call "AI." There are few that understand what these forces mean for the world--and for education and learning--better than Kai-Fu Lee. Lee has been an enormously influential researcher, driving forward work on AI. Originally from Taiwan, he came the U.S. at age 11 and went on to earn degrees from Columbia University and Carnegie Mellon University. He then went on to work at Apple, Microsoft and Google, where he served as president of Google China.

NSF grant to UCLA computer science professors Todd Millstein and Guy Van den Broeck will support research to democratize emerging AI-based technology. Two computer scientists at the UCLA Samueli School of Engineering have received a four-year, $947,000 research grant from the National Science Foundation to make machine learning – a branch of artificial intelligence where computer programs learn and improve on their own – more widely available and easier to work with. "Machine learning has been really successful in the past decade, leading to state-of-the-art techniques for language translation, face recognition and other compelling applications, but these advances have mainly come from experts with specialized knowledge at major technology companies and at universities," said Todd Millstein, professor of computer science and the principal investigator on the research. "Our primary goal with this research is to democratize machine learning, so that programs that utilize it can be written by anyone." Machine learning technologies are powered by probability models.

Two Princeton University computer science professors will lead a new Google AI lab opening in January in the town of Princeton. The lab is expected to expand New Jersey's burgeoning innovation ecosystem by building a collaborative effort to advance research in artificial intelligence. The lab, at 1 Palmer Square, will start with a small number of faculty members, graduate and undergraduate student researchers, recent graduates and software engineers. The lab builds on several years of close collaboration between Google and professors Elad Hazan and Yoram Singer, who will split their time working for Google and Princeton. The work in the lab will focus on a discipline within artificial intelligence known as machine learning, in which computers learn from existing information and develop the ability to draw conclusions and make decisions in new situations that were not in the original data.