A successful machine learning (ML) algorithm often relies on a large amount of high-quality data to train well-performed models. Supervised learning approaches, such as deep learning techniques, generate high-quality ML functions for real-life applications, however with large costs and human efforts to label training data. Recent advancements in federated learning (FL) allow multiple data owners or organisations to collaboratively train a machine learning model without sharing raw data. In this light, vertical FL allows organisations to build a global model when the participating organisations have vertically partitioned data. Further, in the vertical FL setting the participating organisation generally requires fewer resources compared to sharing data directly, enabling lightweight and scalable distributed training solutions. However, privacy protection in vertical FL is challenging due to the communication of intermediate outputs and the gradients of model update. This invites adversary entities to infer other organisations underlying data. Thus, in this paper, we aim to explore how to protect the privacy of individual organisation data in a differential privacy (DP) setting. We run experiments with different real-world datasets and DP budgets. Our experimental results show that a trade-off point needs to be found to achieve a balance between the vertical FL performance and privacy protection in terms of the amount of perturbation noise.

Deep learning models have demonstrated superior performance in various healthcare applications. However, the major limitation of these deep models is usually the lack of high-quality training data due to the private and sensitive nature of this field. In this study, we propose a novel textual data augmentation method to generate artificial clinical notes in patients' Electronic Health Records (EHRs) that can be used as additional training data for patient outcomes prediction. Essentially, we fine-tune the generative language model GPT-2 to synthesize labeled text with the original training data. More specifically, We propose a teacher-student framework where we first pre-train a teacher model on the original data, and then train a student model on the GPT-augmented data under the guidance of the teacher. We evaluate our method on the most common patient outcome, i.e., the 30-day readmission rate. The experimental results show that deep models can improve their predictive performance with the augmented data, indicating the effectiveness of the proposed architecture.

When performing complex tasks, humans naturally reason at multiple temporal and spatial resolutions simultaneously. We contend that for an artificially intelligent agent to effectively model human teammates, i.e., demonstrate computational theory of mind (ToM), it should do the same. In this paper, we present an approach for integrating high and low-resolution spatial and temporal information to predict human behavior in real time and evaluate it on data collected from human subjects performing simulated urban search and rescue (USAR) missions in a Minecraft-based environment. Our model composes neural networks for high and low-resolution feature extraction with a neural network for behavior prediction, with all three networks trained simultaneously. The high-resolution extractor encodes dynamically changing goals robustly by taking as input the Manhattan distance difference between the humans' Minecraft avatars and candidate goals in the environment for the latest few actions, computed from a high-resolution gridworld representation. In contrast, the low-resolution extractor encodes participants' historical behavior using a historical state matrix computed from a low-resolution graph representation. Through supervised learning, our model acquires a robust prior for human behavior prediction, and can effectively deal with long-term observations. Our experimental results demonstrate that our method significantly improves prediction accuracy compared to approaches that only use high-resolution information.

In this paper, we propose a new method for knowledge base completion (KBC): instance-based learning (IBL). For example, to answer (Jill Biden, lived city,? ), instead of going directly to Washington D.C., our goal is to find Joe Biden, who has the same lived city as Jill Biden. Through prototype entities, IBL provides interpretability. We develop theories for modeling prototypes and combining IBL with translational models. Experiments on various tasks confirmed the IBL model's effectiveness and interpretability. In addition, IBL shed light on the mechanism of rule-based KBC models. Previous research has generally agreed that rule-based models provide rules with semantically compatible premises and hypotheses. We challenge this view. We begin by demonstrating that some logical rules represent {\it instance-based equivalence} (i.e. prototypes) rather than semantic compatibility. These are denoted as {\it IBL rules}. Surprisingly, despite occupying only a small portion of the rule space, IBL rules outperform non-IBL rules in all four benchmarks. We use a variety of experiments to demonstrate that rule-based models work because they have the ability to represent instance-based equivalence via IBL rules. The findings provide new insights of how rule-based models work and how to interpret their rules.

Artificial intelligence is rapidly transforming all sectors of our society. Whether we realize it or not, every time we do a Google search or ask Siri a question, we're using AI. For better or worse, the same is true about the very character of warfare. This is the reason why the Department of Defense – like its counterparts in China and Russia– is investing billions of dollars to develop and integrate AI into defense systems. It's also the reason why DoD is now embracing initiatives that envision future technologies, including the next phase of AI – artificial general intelligence.

Amazon has unveiled its latest warehouse robot. It says "Sparrow is the first robotic system in our warehouses that can detect, select, and handle individual products in our inventory." The robotic arm uses AI and computer vision to recognize and handle millions of items, according to Amazon. The company says that, by employing robots in its warehouses, it can conduct operations more efficiently and safely. "Sparrow will take on repetitive tasks, enabling our employees to focus their time and energy on other things, while also advancing safety," Amazon said.

The United States and the European Union plan to release a new artificial intelligence road map that prioritizes security and risk management at the next meeting of their joint trade and technology council, a senior US official said on Thursday. Marisa Lago, commerce undersecretary for international trade, told an event hosted by the US Chamber of Commerce that the document would be released at the next ministerial meeting of the US-EU Trade and Technology Council, on December 5. "We think that this is a mutual priority that is going to grow in scope as new AI applications come online and as more authoritarian regimes are taking a very different approach to the issues of security and risk management," she said. Lago said US and EU officials felt the document would be integral to ensuring that new technologies were deployed in line with shared democratic values and free-market principles. It should also help ensure that small- and medium-sized US and EU businesses are not locked out of new digital markets. US Secretary of Commerce Gina Raimondo met virtually with EU Commission Executive Vice President Margrethe Vestager on Wednesday to discuss TTC work, with a focus on issues such as artificial intelligence, semiconductors and information communication technology services, the Commerce Department said.

Japan is investing almost half a billion dollars to beef up semiconductor development and production in a "last chance" attempt to keep its position as a major player on the global technology stage, the government said Friday. The new company Rapidus, which means "quick" in Latin, will work on developing next-generation, or "post-5G," semiconductors, according to the Ministry of Economy Trade and Industry. These advanced chips will allow for smart gadgets and smart cities with high-speed sensors and transmission. The components have to be extremely thin -- a fraction of a hair's breadth. The 70-billion-yen ($490-million) effort will involve working closely with major Japan ally the U.S. to bring together "the best and the brightest" from both nations, the ministry said in a statement.

Trapped in a metal cage in a corner of a 350,000 sq ft Amazon warehouse outside Boston last week a lonely yellow robot arm sorted through packages, preparing items to be shipped out to customers who demand ever-faster delivery. Soon it will be joined by others in a development that could mean the end of thousands of jobs and, Amazon argues, the creation of thousands of others. As the robot worked, a screen displayed its progress. It carefully packed a tub of protein powder, next came a box of napkin rings then … a tube of hemorrhoid cream. As 100 journalists from around the world snapped pictures, someone switched the screen to hide the cream.

New economic research reveals that teaching Artificial Intelligence (AI) skills in secondary schools could help to fill increasing demand for computer science and AI related roles, supporting on average £71 billion of economic output annually to 2030 in the UK economy. The report – commissioned by Amazon from Capital Economics – estimates that demand for jobs that require computer science, AI or machine learning skills in the UK are expected to increase by 40% over the next five years. In addition, research that looked at the potential future use of AI by UK businesses estimates that expenditure on AI-related labour could increase from £46 billion in 2020 to between £80 billion and £103 billion by 2025. In order to have enough AI talent in the UK workforce to fill computer science jobs by 2030, students will need to experience some form of AI-based learning during secondary school. An insufficient supply of skilled labour is one of the reasons why UK businesses are slow to adopt AI, with just 15% of UK businesses having currently adopted the technology.