In this paper, we propose a model-free reinforcement learning method to synthesize control policies for motion planning problems for continuous states and actions. The robot is modelled as a labeled Markov decision process (MDP) with continuous state and action spaces. Linear temporal logics (LTL) are used to specify high-level tasks. We then train deep neural networks to approximate the value function and policy using an actor-critic reinforcement learning method. The LTL specification is converted into an annotated limit-deterministic B\"uchi automaton (LDBA) for continuously shaping the reward so that dense reward is available during training. A naive way of solving a motion planning problem with LTL specifications using reinforcement learning is to sample a trajectory and, if the trajectory satisfies the entire LTL formula then we assign a high reward for training. However, the sampling complexity needed to find such a trajectory is too high when we have a complex LTL formula for continuous state and action spaces. As a result, it is very unlikely that we get enough reward for training if all sample trajectories start from the initial state in the automata. In this paper, we propose a method that samples not only an initial state from the state space, but also an arbitrary state in the automata at the beginning of each training episode. We test our algorithm in simulation using a car-like robot and find out that our method can learn policies for different working configurations and LTL specifications successfully.

A sum-product network (SPN) is a probabilistic model, based on a rooted acyclic directed graph, in which terminal nodes represent univariate probability distributions and non-terminal nodes represent convex combinations (weighted sums) and products of probability functions. They are closely related to probabilistic graphical models, in particular to Bayesian networks with multiple context-specific independencies. Their main advantage is the possibility of building tractable models from data, i.e., models that can perform several inference tasks in time proportional to the number of links in the graph. They are somewhat similar to neural networks and can address the same kinds of problems, such as image processing and natural language understanding. This paper offers a survey of SPNs, including their definition, the main algorithms for inference and learning from data, the main applications, a brief review of software libraries, and a comparison with related models

Behavioural cloning, where a computer is taught to perform a task based on demonstrations, has been successfully applied to various video games and robotics tasks, with and without reinforcement learning. This also includes end-to-end approaches, where a computer plays a video game like humans do: by looking at the image displayed on the screen, and sending keystrokes to the game. As a general approach to playing video games, this has many inviting properties: no need for specialized modifications to the game, no lengthy training sessions and the ability to re-use the same tools across different games. However, related work includes game-specific engineering to achieve the results. We take a step towards a general approach and study the general applicability of behavioural cloning on twelve video games, including six modern video games (published after 2010), by using human demonstrations as training data. Our results show that these agents cannot match humans in raw performance but can learn human-like behaviour. We also demonstrate how the quality of the data matters, and how recording data from humans is subject to a state-action mismatch, due to human reflexes.

Disclaimer: The following is based on my observations of machine learning teams -- not an academic survey of the industry. For years, the biggest bottleneck to production deep learning was simple: we needed models that worked. And over the last decade--thanks to companies with access to unprecedented amounts of data and computer power, as well as new model architectures--we've largely cleared that hurdle. We may not have fully autonomous vehicles or Bladerunner-esque AI, but when you call an Uber, you get an accurate ETA prediction. When you open an email in Gmail, you get a contextually appropriate suggestion from Smart Compose.

As global energy regulations are strengthened, improving energy efficiency while maintaining performance of electronic appliances is becoming more important. Especially in air conditioning, energy efficiency can be maximized by adaptively controlling the airflow based on detected human locations; however, several limitations such as detection areas, the installation environment, and sensor quantity and real-time performance which come from the constraints in the embedded system make it a challenging problem. In this study, by using a low resolution cost effective vision sensor, the environmental information of living spaces and the real-time locations of humans are learned through a deep learning algorithm to identify the living area from the entire indoor space. Based on this information, we improve the performance and the energy efficiency of air conditioner by smartly controlling the airflow on the identified living area. In experiments, our deep learning based spatial classification algorithm shows error less than 5 .

From its epicenter in China, the novel coronavirus has spread to infect 414,179 people and cause no less than 18,440 deaths in at least 160 countries across a three-month span from January 2020 till date. These figures are according to the World Health Organization (WHO) Situation report as of March 25th. Accompanying the tragic loss of life that the virus has caused is the impact to the global economy, which has reeled from the effects of the pandemic. Due to the lockdown measures imposed by several governments, economic activity has slowed around the world, and the Organization for Economic Cooperation and Development (OECD) has stated that the global economy could be hit by its worst growth rate since 2009. The OECD have alerted that the growth rate could be as slow as 2.4%, potentially dragging many countries into recession.

When I reviewed Amazon SageMaker in 2018, I noted that it was a highly scalable machine learning and deep learning service that supports 11 algorithms of its own, plus any others you supply. Hyperparameter optimization was still in preview, and you needed to do your own ETL and feature engineering. Since then, the scope of SageMaker has expanded, augmenting the core notebooks with IDEs (SageMaker Studio) and automated machine learning (SageMaker Autopilot) and adding a bunch of important services to the overall ecosystem, as shown in the diagram below. This ecosystem supports machine learning from preparation through model building, training, and tuning to deployment and management -- in other words, end to end. Amazon SageMaker Studio improves on the older SageMaker notebooks, and a number of new services have enhanced the SageMaker ecosystem to support end-to-end machine learning.

In a preprint paper published this week by DeepMind, Google parent company Alphabet's U.K.-based research division, a team of scientists describe Agent57, which they say is the first system that outperforms humans on all 57 Atari games in the Arcade Learning Environment data set. Assuming the claim holds water, Agent57 could lay the groundwork for more capable AI decision-making models than have been previously released. This could be a boon for enterprises looking to boost productivity through workplace automation; imagine AI that automatically completes not only mundane, repetitive tasks like data entry, but which reasons about its environment. "With Agent57, we have succeeded in building a more generally intelligent agent that has above-human performance on all tasks in the Atari57 benchmark," wrote the study's coauthors. "Agent57 was able to scale with increasing amounts of computation: the longer it trained, the higher its score got."

The integration of Azure Machine Learning and Azure IoT Edge enables organizations and developers to apply AI and ML to data that can't make it to the cloud due to data sovereignty, privacy, and/or bandwidth issues. All models created using Azure Machine Learning can now be deployed to IoT gateways and devices with the Azure IoT Edge runtime. Models are operationalized as containers and can run on many types of hardware, from very small devices all the way to powerful servers. We're releasing this toolkit to help get you started with AI and Azure IoT Edge. The toolkit will show you how to package deep learning models in Azure IoT Edge-compatible Docker containers and expose those models as REST APIs.

OneConnect, a leading technology-as-a-service platform serving financial institutions in China, is pleased to announce that its artificial intelligence research institute, Gamma Lab, won the FinTech Team of the Year award for its strong technical prowess, wide range of deployment scenarios across the financial sector and high-speed growth at The Asset Triple A Digital Awards 2020 held by international authoritative media The Asset. The Gamma O platform was awarded the Best Digital Financial Project for its success since launch in providing one-stop solutions that empowered financial institutions and technology service providers in connecting with each other. The Asset was founded in 1999, with its Triple A awards gaining a high level of influence and authority in Asian and international financial markets. For two consecutive years, Gamma Lab won the FinTech Team of the Year award, demonstrating OneConnect's industry leading position in both AI technology R&D and deployment. OneConnect's information extraction technology led at the international AI competition SemEval 2020, representing another world first for Gamma Lab in new AI technologies beyond the successes that the institute had achieved in terms of performance in the areas of microexpression recognition, facial action unit recognition, machine reading comprehension, natural language generation, emotion recognition and deep learning model inference.