Dynamic Bayesian networks have been applied widely to reconstruct the structure of regulatory processes from time series data. The standard approach is based on the assumption of a homogeneous Markov chain, which is not valid in many real-world scenarios. Recent research efforts addressing this shortcoming have considered undirected graphs, directed graphs for discretized data, or over-flexible models that lack any information sharing between time series segments. In the present article, we propose a non-stationary dynamic Bayesian network for continuous data, in which parameters are allowed to vary between segments, and in which a common network structure provides essential information sharing across segments. Our model is based on a Bayesian change-point process, and we apply a variant of the allocation sampler of Nobile and Fearnside to infer the number and location of the change-points.

If recent television shows are anything to go by, we're a little concerned about the consequences of technological development. Black Mirror projects the negative consequences of social media, while artificial intelligence turns rogue in The 100 and Better Than Us. The potential extinction of the human race is up for grabs in Travellers, and Altered Carbon frets over the separation of human consciousness from the body. And Humans and Westworld see trouble ahead for human-android relations. Narratives like these have a long lineage.

In extreme classification settings, embedding-based neural network models are currently not competitive with sparse linear and tree-based methods in terms of accuracy. Most prior works attribute this poor performance to the low-dimensional bottleneck in embedding-based methods. In this paper, we demonstrate that theoretically there is no limitation to using low-dimensional embedding-based methods, and provide experimental evidence that overfitting is the root cause of the poor performance of embedding-based methods. These findings motivate us to investigate novel data augmentation and regularization techniques to mitigate overfitting. To this end, we propose GLaS, a new regularizer for embedding-based neural network approaches.

The European Commission says that the EU could become the most attractive, secure and dynamic data-agile economy in the world. The Commission's new data strategy is for the EU to seize new opportunities in digitised industry and business-to-business artificial intelligence (AI) applications. However, the Commission has scrupulously avoided the vital question of whether GDPR is an obstacle to the EU's plans to become an AI hub. The European Commission announced its new EU data strategy with the publication of two papers in February 2020. These were a white paper on AI and a communication entitled, "A European strategy for data". The Commission acknowledges that "the availability of data is essential for training artificial intelligence systems … without data, there is no AI."

In most large organizations, artificial intelligence (AI) and machine learning (ML) are powering key business functions, from big data analytics and customer service to fraud detection and personalized marketing. Insights that AI and ML can produce are powerful, but it's often difficult, if not impossible, to explain how these algorithms arrived at them. This limitation could pose significant problems for compliance with the EU's General Data Protection Regulation (GDPR), the California Consumer Privacy Act (CCPA), and other laws governing data and privacy. Let's look at GDPR first. When an automated process such as AI or ML makes a decision about an individual based on personal data, GDPR requires the organization to supply an explanation if requested.

Over the last few years, artificial intelligence (AI) has been the enfant terrible of the business world: a technology full of unconventional and sometimes controversial behaviour that has shocked, provoked and enchanted audiences worldwide. But now it's time for AI to grow up. Businesses and consumers are tired of having the same debates around the hype vs reality of AI. In 2020, I see three opportunities for this to happen across responsibility, advocacy and regulation. As AI becomes more pervasive, we're likely to see those wronged by it inspired to take action.

This legislation would accelerate and coordinate Federal investments and facilitate new public-private partnerships in research, standards, and education in artificial intelligence, in order to ensure the United States leads the world in the development and use of responsible artificial intelligence systems. "The United States must act now to cement our global leadership in artificial intelligence and ensure the development and adoption of trustworthy AI systems," said Chairwoman Johnson. "We can accomplish this by accelerating our investments in research, development, and the education and training of an AI workforce, all governed by principles of ethics, safety, security, fairness, and transparency. That is exactly what H.R. 6216 seeks to do. I want to thank Ranking Member Lucas for joining me in the development and introduction of this important legislation and the many stakeholders who advised us during its development."

Enrico Bonadio is Reader at The City Law School, where he teaches various modules on intellectual property (IP) law. He holds law degrees from the University of Florence (PHD) and the University of Pisa (LLB), and is Associate Editor and Intellectual Property Correspondent of the European Journal of Risk Regulation as well as a member of the Editorial Board of NUART Journal. Enrico is also researching on IP protection of AI and robotics: he is part of a consortium that has been awarded funding by the EU as part of Horizon2020 to assess the area of interactive robots in society (INBOTS project). He also recently authored a report on Standard Essential Patents and the Internet of Things (commissioned by the European Parliament). Enrico has been awarded grant funding for other projects, including substantial grants from the ESRC, HEIF, the UK Global Challenges Research Fund and the Australian Research Council.

In light of International Women's Day celebrations this past weekend, we acknowledged the beauty, essence and power of women to achieve and thrive in the global ecosystem. Yet in our modern digital age, women continue to be neglected on multiple fronts, especially that of the new workforce. It is society's role to ensure that all females are given equal opportunities to grow in this new age workforce, and we must understand that all of us have a stake in this mission. Women are the key piece to the puzzle of realizing the highest maturity levels of digital enterprises, but unless we realize this, our progress in AI and technology will remain stagnant. In order to close the gender gap in science, technology, engineering and math (STEM), and to accelerate advances in artificial intelligence and the sciences, we must encourage and support women on all levels, from government to enterprise, and establish equal employment opportunities for all. AI is one of the fields in which women can experience tremendous success, especially with the right push towards female participation in the industry.

We describe nonparametric deconvolution models (NDMs), a family of Bayesian nonparametric models for collections of data in which each observation is the average over the features from heterogeneous particles. For example, these types of data are found in elections, where we observe precinct-level vote tallies (observations) of individual citizens' votes (particles) across each of the candidates or ballot measures (features), where each voter is part of a specific voter cohort or demographic (factor). Like the hierarchical Dirichlet process, NDMs rely on two tiers of Dirichlet processes to explain the data with an unknown number of latent factors; each observation is modeled as a weighted average of these latent factors. Unlike existing models, NDMs recover how factor distributions vary locally for each observation. This uniquely allows NDMs both to deconvolve each observation into its constituent factors, and also to describe how the factor distributions specific to each observation vary across observations and deviate from the corresponding global factors. We present variational inference techniques for this family of models and study its performance on simulated data and voting data from California. We show that including local factors improves estimates of global factors and provides a novel scaffold for exploring data.