A simple yet effective architectural design of radial basis function neural networks (RBFNN) makes them amongst the most popular conventional neural networks. The current generation of radial basis function neural network is equipped with multiple kernels which provide significant performance benefits compared to the previous generation using only a single kernel. In existing multi-kernel RBF algorithms, multi-kernel is formed by the convex combination of the base/primary kernels. In this paper, we propose a novel multi-kernel RBFNN in which every base kernel has its own (local) weight. This novel flexibility in the network provides better performance such as faster convergence rate, better local minima and resilience against stucking in poor local minima. These performance gains are achieved at a competitive computational complexity compared to the contemporary multi-kernel RBF algorithms. The proposed algorithm is thoroughly analysed for performance gain using mathematical and graphical illustrations and also evaluated on three different types of problems namely: (i) pattern classification, (ii) system identification and (iii) function approximation. Empirical results clearly show the superiority of the proposed algorithm compared to the existing state-of-the-art multi-kernel approaches.

To explore heterogeneous information networks (HINs), network representation learning (NRL) is proposed, which represents a network in a low-dimension space. Recently, graph neural networks (GNNs) have drawn a lot of attention which are very expressive for mining a HIN, while they suffer from low efficiency issue. In this paper, we propose a pre-training and fine-tuning framework PF-HIN to capture the features of a HIN. Unlike traditional GNNs that have to train the whole model for each downstream task, PF-HIN only needs to fine-tune the model using the pre-trained parameters and minimal extra task-specific parameters, thus improving the model efficiency and effectiveness. Specifically, in pre-training phase, we first use a ranking-based BFS strategy to form the input node sequence. Then inspired by BERT, we adopt deep bi-directional transformer encoders to train the model, which is a variant of GNN aggregator that is more powerful than traditional deep neural networks like CNN and LSTM. The model is pre-trained based on two tasks, i.e., masked node modeling (MNM) and adjacent node prediction (ANP). Additionally, we leverage factorized embedding parameterization and cross-layer parameter sharing to reduce the parameters. In fine-tuning stage, we choose four benchmark downstream tasks, i.e., link prediction, similarity search, node classification and node clustering. We use node sequence pairs as input for link prediction and similarity search, and a single node sequence as input for node classification and clustering. The experimental results of the above tasks on four real-world datasets verify the advancement of PF-HIN, as it outperforms state-of-the-art alternatives consistently and significantly.

With the increasing adoption of Deep Learning (DL) for critical tasks, such as autonomous driving, the evaluation of the quality of systems that rely on DL has become crucial. Once trained, DL systems produce an output for any arbitrary numeric vector provided as input, regardless of whether it is within or outside the validity domain of the system under test. Hence, the quality of such systems is determined by the intersection between their validity domain and the regions where their outputs exhibit a misbehaviour. In this paper, we introduce the notion of frontier of behaviours, i.e., the inputs at which the DL system starts to misbehave. If the frontier of misbehaviours is outside the validity domain of the system, the quality check is passed. Otherwise, the inputs at the intersection represent quality deficiencies of the system. We developed DeepJanus, a search-based tool that generates frontier inputs for DL systems. The experimental results obtained for the lane keeping component of a self-driving car show that the frontier of a well trained system contains almost exclusively unrealistic roads that violate the best practices of civil engineering, while the frontier of a poorly trained one includes many valid inputs that point to serious deficiencies of the system.

This week we speak to Andrew Day, chief executive of computer games developer Keywords Studios. Andrew Day knows from personal experience just how addictive some computer games can be. "I have one of those horrible personalities, that if I open a game, I find, before I know where I am, that I have spent tens of hours on it," says the 56-year-old. I went for a little break, I was lying beside a swimming pool with nothing to do. So I picked up my phone and started playing Candy Crush.

COVID-19 sparked an unprecedented global health crisis around the world and ushered us to a'new normal' way of working from home almost overnight. Not all companies were ready to adapt to this unexpected disruption, and business suffered. Many organisations faced massive structural changes and looked at alternative business strategies to sustain themselves through this global pandemic. But there is always an upside to every crisis – innovation leaders such as SAP have introduced innovative solutions specifically designed to help businesses in the post-COVID-19 global economy. We are a long way from business as usual these days, as many of us juggle work and home responsibilities, having video conferences interrupted by our kids, dogs barking and kitchen appliances whirring in the background.

Your smartphone knows when you are at home or the office. At least, mine does, and can even tell me when to leave to get at one of my common destinations on time. We all accept that our smart devices collect information about our preferences and send them over to the cloud for processing. These come back as recommendations for shopping, food, mating, and when to leave the office and head home. What is the magic behind inferring a usual location?

The first day of The Rising 2020 started with an informal session with Sara Hooker, a researcher at Google Brain where she shared some of her personal reflections on how to navigate in the field of machine learning and why we need to celebrate failures as well as success. Sara started her session with a simple story where she shared her childhood dream of being featured in the magazine, The Economist. In fact, she mentioned that "one of my goals was to eventually be an economist." However, when that happened in 2016, it wasn't a pleasing feeling for Sara; instead, it was a feeling of "unease" and seemed problematic. A lot of this could be attributed to the article that The Economist did, which profiled the efforts of fast.ai, a course that's run by Jeremy Howard and Rachel Thomas, and utilised Sara as an example of their success.

One of NVIDIA's many different artificial intelligence projects (and by far the best one to date) lets you envision what your pet might look like it it were a meerkat. In case you didn't know, NVIDIA has its own research group dedicated solely to research into AI, and that includes developing new AI systems and agents which can do some pretty neat things. As the researchers say, although they take AI research very seriously, there's still no excuse not to have some fun with the products of their labors. It's the name given to an AI system they developed around a year ago which can generate a selection of images that are sorts of translations of your own pet's face into what said pet might look like if they were other types of animals. "With GANimal, you can bring your pet's alter ego to life by projecting their expression and pose onto other animals," explain the developers.

Each Fourth of July for the past five years I've written about AI with the potential to positively impact democratic societies. I return to this question with the hope of shining a light on technology that can strengthen communities, protect privacy and freedoms, or otherwise support the public good. This series is grounded in the principle that artificial intelligence can is capable of not just value extraction, but individual and societal empowerment. While AI solutions often propagate bias, they can also be used to detect that bias. As Dr. Safiya Noble has pointed out, artificial intelligence is one of the critical human rights issues of our lifetimes.

A smartphone that can warn you not to send a text while you're upset? Early in my career--back in the stone age before computers and smartphones--I worked in environments where memos were a primary means of communication. Sure, my colleagues and I could talk face-to-face, but the culture of the time was to memorialize much of our interaction in writing. Believe it or not, there were some advantages in what now seems such an archaic practice. Unlike texts and emails--where one tap of the "send" button can fill you with instant regret--the old-fashioned memo provided a cushion of safety, a chance to reconsider.