Abstaining classifiers have been widely used in cost-sensitive applications to avoid ambiguous classification and reduce the cost of misclassification. Previous abstaining classification models rely on cost information, such as a cost matrix or cost ratio. However, it is difficult to obtain or estimate costs in practical applications. Furthermore, these abstention models are typically restricted to a single optimization metric, which may not be the expected indicator when evaluating classification performance. To overcome such problems, a multi-objective bounded-abstention (MOBA) model is proposed to optimize essential metrics. Specifically, the MOBA model minimizes the error rate of each class under class-dependent abstention constraints. The MOBA model is then solved using the non-dominated sorting genetic algorithm II, which is a popular evolutionary multi-objective optimization algorithm. A set of Pareto-optimal solutions will be generated and the best one can be selected according to provided conditions (whether costs are known) or performance demands (e.g., obtaining a high accuracy, F-measure, and etc). Hence, the MOBA model is robust towards variations in the conditions and requirements. Compared to state-of-the-art abstention models, MOBA achieves lower expected costs when cost information is considered, and better performance-abstention trade-offs when it is not.

Off-policy reinforcement learning with eligibility traces is challenging because of the discrepancy between target policy and behavior policy. One common approach is to measure the difference between two policies in a probabilistic way, such as importance sampling and tree-backup. However, existing off-policy learning methods based on probabilistic policy measurement are inefficient when utilizing traces under a greedy target policy, which is ineffective for control problems. The traces are cut immediately when a non-greedy action is taken, which may lose the advantage of eligibility traces and slow down the learning process. Alternatively, some non-probabilistic measurement methods such as General Q($\lambda$) and Naive Q($\lambda$) never cut traces, but face convergence problems in practice. To address the above issues, this paper introduces a new method named TBQ($\sigma$), which effectively unifies the tree-backup algorithm and Naive Q($\lambda$). By introducing a new parameter $\sigma$ to illustrate the \emph{degree} of utilizing traces, TBQ($\sigma$) creates an effective integration of TB($\lambda$) and Naive Q($\lambda$) and continuous role shift between them. The contraction property of TB($\sigma$) is theoretically analyzed for both policy evaluation and control settings. We also derive the online version of TBQ($\sigma$) and give the convergence proof. We empirically show that, for $\epsilon\in(0,1]$ in $\epsilon$-greedy policies, there exists some degree of utilizing traces for $\lambda\in[0,1]$, which can improve the efficiency in trace utilization for off-policy reinforcement learning, to both accelerate the learning process and improve the performance.

A decentralized learning mechanism, Federated Learning (FL), has attracted much attention, which enables privacy-preserving training using the rich data and computational resources of mobile clients. However, data on mobile clients is typically not independent and identically distributed (IID) owing to diverse of mobile users' interest and usage, and FL on non-IID data could degrade the model performance. This work aims to extend FL to solve the performance degradation problem resulting from non-IID data of mobile clients. We assume that a limited number (e.g., less than 1%) of clients who allow their data to be uploaded to a server, and we propose a novel learning mechanism referred to as Hybrid-FL, where the server updates the model using data gathered from the clients and merge the model with models trained by clients. In Hybrid-FL, we design a heuristic algorithms that solves the data and client selection problem to construct "good" dataset on the server under bandwidth and time limitation. The algorithm increases the amount of data gathered from clients and makes the data approximately IID for improving model performance. Evaluations consisting of network simulations and machine learning (ML) experiments show that the proposed scheme achieves a significantly higher classification accuracy than previous schemes in the non-IID case.

We design a self size-estimating feed-forward network (SSFN) using a joint optimization approach for estimation of number of layers, number of nodes and learning of weight matrices at a low computational complexity. In the proposed approach, SSFN grows from a small-size network to a large-size network. The increase in size from small-size to large-size guarantees a monotonically decreasing cost with addition of nodes and layers. The optimization approach uses a sequence of layer-wise target-seeking non-convex optimization problems. Using `lossless flow property' of some activation functions, such as rectified linear unit (ReLU), we analytically find regularization parameters in the layer-wise non-convex optimization problems. Closed-form analytic expressions of regularization parameters allow to avoid tedious cross-validations. The layer-wise non-convex optimization problems are further relaxed to convex optimization problems for ease of implementation and analytical tractability. The convex relaxation helps to design a low-complexity algorithm for construction of the SSFN. We experiment with eight popular benchmark datasets for sound and image classification tasks. Using extensive experiments we show that the SSFN can self-estimate its size using the low-complexity algorithm. The size of SSFN varies significantly across the eight datasets.

An advanced research arm of the U.S. government's intelligence community is looking to develop AI capable of tracking people across a vast surveillance network. As reported by Nextgov, the Intelligence Advanced Research Projects Activity (IARPA) has put out a call for more information on developing an algorithm that can be trained to identify targets by visually analyzing swaths of security camera footage. The goal, says the request, is to be able to identify and track subjects across areas as large as six miles in an effort to reconstruct crime scenes, protect military operations, and monitor critical infrastructure facilities. To develop the technology, IARPA will collect nearly 1,000 hours of video surveillance from at least 20 camera networks and then, using that sample, test various algorithms effectiveness. The agency's interest in AI-based surveillance technology mirrors a broader movement from governments and intelligence communities around the globe, many of whom have ramped up efforts to develop and scale systems.

Organizers of the 2020 Tokyo Olympic and Paralympic Games announced plans Wednesday to launch robots from the "Mobile Suit Gundam" anime series into space aboard a satellite that will broadcast messages of support to athletes. In the project, conducted in collaboration with the Japan Aerospace Exploration Agency and the University of Tokyo, two 10-centimeter models depicting Gundam and Char's Zaku robots from the animation series will be sent into orbit on a 30-cm long, 10-cm wide microsatellite. The "G Satellite," with an electronic bulletin board for displaying messages, will be sent to the International Space Station aboard a supply ship next March and later launched from the ISS. After the satellite enters the Earth's orbit, it will deploy the robots and the bulletin board. The organizers of the project will then share images taken with an onboard camera, including congratulatory messages in multiple languages, with athletes through social media and other outlets.

Artificial Intelligence is a constantly transforming industry and is rapidly becoming a profitable one as well with many opportunities to invest at the ground level this year with potential for tremendous growth. Here are 5 AI companies you should investigate investing stock in this year. This is one of the largest social media companies to come out of China. They recently founded an AI lab which developed tools to process information across its ecosystem. It developed tools for natural language processing, news aggregators and facial recognition.

AI and People Analytics have taken off. As I've written about in the past, the workplace has become a highly instrumented place. Companies use surveys and feedback tools to get our opinions, new tools monitor emails and our network of communications (ONA), we capture data on travel, location, and mobility, and organizations now have data on our wellbeing, fitness, and health. And added to this is a new stream of data which includes video (every video conference can be recorded and more than 40% of job interviews are recorded), audio (tools that record meetings can sense mood), and image recognition that recognizes faces wherever we are. In the early days of HR analytics, companies captured employee data to measure span of control, the distribution of performance ratings, succession pipeline, and other talent-related topics. Today, with all this new information entering the workplace (virtually everywhere you click at work is stored somewhere), the domain of people analytics is getting very personal. While I know HR professionals take the job of ethics and safety seriously, I'd like to point out some ethical issues we need to consider.

The HKBU team, entitled AI Phoenix, included team leader and mathematics alumnus Xu Zhouming; computer science students Lyu Jiayou, Zeng Xuan, Wang Shihao and Xu Chen, as well as mathematics alumna Xu Fangfei. The HKBU team proposed the application of artificial intelligence to customer relationship management in the banking industry. The team enhanced the architecture of AI engines, making financial market data analysis more effective, on top to the current image processing and speech recognition applications. The enhancements are intended to raise the retail banks' competitiveness through more effective analysis of customer attributes. Team leader Xu Zhouming said a bank on the Mainland will explore a more in-depth collaboration with AI Phoenix Technology Company Limited, the start-up established by the team.

The future impacts of artificial intelligence (AI) on society and the labour force have been studied and reported extensively. In a recent book, AI Superpowers, Kai-Fu Lee, former president of Google China, wrote that 40 to 50 per cent of current jobs will be technically and economically viable with AI and automation over the next 15 years. Artificial intelligence refers to computer systems that collect, interpret and learn from external data to achieve specific goals and tasks. Unlike natural intelligence displayed by humans and animals, it is an artificial form of intelligence demonstrated by machines. This has raised questions about the ethics of AI decision-making and impacts of AI in the workplace.