We present in this work a fully Transformer-based reinforcement learning generator-evaluator architecture for neural question generation. Question generation is a task that consists in generating questions given a context and answer. To improve the quality of the generated question, we came up with a semantic-based self-critical training layout in generator-evaluator architecture, which goes beyond typical maximum likelihood training. Evaluation metrics for language modeling only based on n-gram overlapping do not consider semantic relations between reference and candidate strings. To improve the evaluation step, we assess our model for both n-gram overlap using BLEU and semantically using BERTScore and NUBIA, a novel state-of-the-art evaluation metric for text generation. Question generation could be used in many downstream applications, including in extending question answering datasets, conversational systems, and educational assessment systems.

Achieving human-level performance on some of Machine Reading Comprehension (MRC) datasets is no longer challenging with the help of powerful Pre-trained Language Models (PLMs). However, the internal mechanism of these artifacts still remains unclear, placing an obstacle for further understanding these models. This paper focuses on conducting a series of analytical experiments to examine the relations between the multi-head self-attention and the final performance, trying to analyze the potential explainability in PLM-based MRC models. We perform quantitative analyses on SQuAD (English) and CMRC 2018 (Chinese), two span-extraction MRC datasets, on top of BERT, ALBERT, and ELECTRA in various aspects. We discover that {\em passage-to-question} and {\em passage understanding} attentions are the most important ones, showing strong correlations to the final performance than other parts. Through visualizations and case studies, we also observe several general findings on the attention maps, which could be helpful to understand how these models solve the questions.

Item Response Theory (IRT) is a ubiquitous model for understanding human behaviors and attitudes based on their responses to questions. Large modern datasets offer opportunities to capture more nuances in human behavior, potentially improving psychometric modeling leading to improved scientific understanding and public policy. However, while larger datasets allow for more flexible approaches, many contemporary algorithms for fitting IRT models may also have massive computational demands that forbid real-world application. To address this bottleneck, we introduce a variational Bayesian inference algorithm for IRT, and show that it is fast and scalable without sacrificing accuracy. Applying this method to five large-scale item response datasets from cognitive science and education yields higher log likelihoods and higher accuracy in imputing missing data than alternative inference algorithms. Using this new inference approach we then generalize IRT with expressive Bayesian models of responses, leveraging recent advances in deep learning to capture nonlinear item characteristic curves (ICC) with neural networks. Using an eigth-grade mathematics test from TIMSS, we show our nonlinear IRT models can capture interesting asymmetric ICCs. The algorithm implementation is open-source, and easily usable.

A full-time job in Machine learning can be overwhelming. Balancing a 9–5 job and keeping out of work hours learning routine was more manageable at the start of my career than now. The excitement and novelty of the machine learning field, coupled with a role with few responsibilities, meant that I could spend an extra two hours outside of work staying up to date with ML developments and complete personal projects. But as the months went by, my responsibilities and workload increased. Gaining more responsibility within your ML role is an achievement (especially if accompanied with).

JMU College Republicans chairwoman Juliana McGrath shares her concerns with controversial diversity, equity and inclusion training for first-year students. James Madison University is under fire for pushing controversial rhetoric as part of its freshmen orientation training for student leaders. The PowerPoint presentation and accompanying video addressed topics like social justice, identity, power and privilege, and labeled any person who fits the parameters of White, male, straight and Christian as oppressors in a detailed chart. JMU College Republicans chairwoman Juliana McGrath shared her frustration with Fox News, saying the training at the Virginia university that's meant to bring students together will ultimately be divisive. "When you're teaching about things like this, the goal is to try to bring people together and try to get people to understand different life experiences," she said.

Nine in 10 UK employees will have to reskill by 2030, according to a report based on analysis by McKinsey, but where should companies start? Whether you call it artificial intelligence, machine learning or automation, new media articles appear on the subject daily to spread fear about workers losing jobs as'robots take over'. There's no doubt that the world of work is changing fast, but what it really indicates is our need to adapt to yet another industrial revolution -- the fourth one. In a hyper-technical world, having a high EQ will be more important than ever. This being the case, we've got no time to waste when it comes to upskilling workers.

Python is one of the most popular programming languages. It is mostly preferred by data scientists, machine learning engineers, and web developers. Learning the basics of Python properly is very important because once you have learned the basics, you can learn Python for any profession. So, if you are looking for the best courses to learn Python, this article is for you. In this article, I'm going to introduce you to some of the best Python courses that will help you learn Python.

Recent research efforts in lifelong learning propose to grow a mixture of models to adapt to an increasing number of tasks. The proposed methodology shows promising results in overcoming catastrophic forgetting. However, the theory behind these successful models is still not well understood. In this paper, we perform the theoretical analysis for lifelong learning models by deriving the risk bounds based on the discrepancy distance between the probabilistic representation of data generated by the model and that corresponding to the target dataset. Inspired by the theoretical analysis, we introduce a new lifelong learning approach, namely the Lifelong Infinite Mixture (LIMix) model, which can automatically expand its network architectures or choose an appropriate component to adapt its parameters for learning a new task, while preserving its previously learnt information. We propose to incorporate the knowledge by means of Dirichlet processes by using a gating mechanism which computes the dependence between the knowledge learnt previously and stored in each component, and a new set of data. Besides, we train a compact Student model which can accumulate cross-domain representations over time and make quick inferences. The code is available at https://github.com/dtuzi123/Lifelong-infinite-mixture-model.

This past year has seen a significant blossoming of discussions on the ethics of AI. In working groups and meetings spanning IEEE, ACM, U.N. and the World Economic Forum as well as a handful of governmental advisory committees, more intimate breakout sessions afford an opportunity to observe how we, as robotics and AI researchers, communicate our own relationship to ethics within a field teeming with possibilities of both benefit and harm. Unfortunately, many of these opportunities fail to realize authentic forward progress during discussions that repeat similar memes. Three common myths pervade such discussions, frequently stifling any synthesis: education is not needed; external regulation is undesirable; and technological optimism provides justifiable hope. The underlying good news is that discourse and curricular experimentation are now occurring at scales that were unmatched in the recent past.

Lamont A. Flowers (lflower@clemson.edu) is the Distinguished Professor of Educational Leadership in the Department of Educational and Organizational Leadership Development in the College of Education and the Executive Director of the Charles H. Houston Center for the Study of the Black Experience in Education in the Division of Inclusion and Equity at Clemson University, Clemson, SC, USA.