It's no secret that Disney films of yore had been plagued with racism and sexism. The company's movies have gotten more progressive in recent years --see Brave, Frozen, and Moana--but there's still lots more work to do. And now, Disney has pledged to tackle diversity in storytelling and gender bias with a little help from AI. The company has teamed up with Geena Davis and her Institute on Gender in Media to use GD-IQ: Spellcheck for Bias (Geena Davis Inclusion Quotient), a tool that analyzes TV and movie scripts to track gender and other biases. The software, codeveloped by the University of Southern California Viterbi School of Engineering, evaluates the number of male and female characters, how many characters are part of the LGBTQIA community, how many people of color are included, and how many disabled people are represented.

Conversational interfaces have gone mainstream. The technology behind keeps crossing new milestones, the result of which chatbots have transformed from simple Q&A systems to intelligent personal assistants. As a result, bots found widespread application in diverse areas, most recently in education. Although education stayed backward in terms of technology adoption, lately it took on a renewed quest to incorporate it. Educators are on the lookout for innovative ed-tech systems for efficient tutoring and students increasingly prefer personalized learning environments.

Amongst those who have, 82% answered the question they were asked. CLEARD LIFE RESPONSE: It's good to know the difference between Recruiters and the Cleard.life We can ask, in order to assess Trustworthiness and Loyalty. We can ask, in order to evaluate the potential for coercion, manipulation – assessing Trustworthiness, Maturity and Loyalty. We can ask in order to (think adult children involved in illegal conduct or having criminal associations placing the Candidate/Employer in harm) assess Trustworthiness and Loyalty.

Experts from Australia and around the world in the humanities, social and technological sciences are collaborating on a government-funded initiative to investigate how automated decision-making technologies can be used safely and ethically.

Nearly four in five (79%) Australian IT leaders believe that AI will be very or critically important to their business within two years, but only 34% have a comprehensive, company-wide AI strategy. These are among the key findings of new research by Deloitte into the state of AI in the enterprise in seven markets including Australia. According to the report, more than half (51%) of Australian respondents believe that AI will transform their business within three years. But the survey found that Australia is below the global average both in terms of the proportion of companies with a comprehensive AI strategy (35% globally) and in terms of the percentage of companies that are seasoned AI adopters (17%, compared to 21% globally). In addition, 50% of AI early adopters in Australia are still using AI to catch up or keep up with the competition rather than carve out a distinct advantage.

Rapidly maturing technologies such as 5G, artificial intelligence (AI), and quantum computing will "shift fundamentally the global balance of power", according to Dr Tobias Feakin, Australia's Ambassador for Cyber Affairs. "Those [nations] that really are at the forefront of AI and the way that it works will genuinely be at the forefront of the emerging 21st century economy," he told the Australian Cybersecurity Conference, or CyberCon, in Melbourne on Wednesday. Some nations are already positioning themselves to take advantage of these technologies. "Geopolitics now is being shaped and harnessed in a way that we probably didn't think conceivable a decade ago," Feakin said. "We need to be thinking about grand strategy in technology. How do we ensure that we're plugging into this conversation and the kinds of areas that are shaping technology, not only the technology development itself, but the kinds of legislation that shape the absorption of that technology into the global environment," he said.

For the first time ever, a human drug has been created entirely by artificial intelligence (AI). This news comes from a team at Flinders University in Australia, who claims to have created an enhanced influenza vaccine using an AI program known Search Algorithm for Ligands (SAM). Though computers have been used to make drugs before, this was the first time it was done independently by an AI system. The researchers described this drug as a flu vaccine with an added compound that better stimulates the human immune system. This addition causes more antibodies to be formed against the flu virus than with the traditional vaccination, increasing the vaccine's efficacy.

In the domain of machine learning, Neural Memory Networks (NMNs) have recently achieved impressive results in a variety of application areas including visual question answering, trajectory prediction, object tracking, and language modelling. However, we observe that the attention based knowledge retrieval mechanisms used in current NMNs restricts them from achieving their full potential as the attention process retrieves information based on a set of static connection weights. This is suboptimal in a setting where there are vast differences among samples in the data domain; such as anomaly detection where there is no consistent criteria for what constitutes an anomaly. In this paper, we propose a plastic neural memory access mechanism which exploits both static and dynamic connection weights in the memory read, write and output generation procedures. We demonstrate the effectiveness and flexibility of the proposed memory model in three challenging anomaly detection tasks in the medical domain: abnormal EEG identification, MRI tumour type classification and schizophrenia risk detection in children. In all settings, the proposed approach outperforms the current state-of-the-art. Furthermore, we perform an in-depth analysis demonstrating the utility of neural plasticity for the knowledge retrieval process and provide evidence on how the proposed memory model generates sparse yet informative memory outputs.

ABSTRACT There is an implicit assumption that traditional hybrid approaches for automatic speech recognition (ASR) cannot directly model graphemes and need to rely on phonetic lexicons to get competitive performance, especially on English which has poor grapheme-phoneme correspondence. In this work, we show for the first time that, on English, hybrid ASR systems can in fact model graphemes effectively by leveraging tied context-dependent graphemes, i.e., chenones. Our chenone-based systems significantly outperform equivalent senone baselines by 4.5% to 11.1% relative on three different English datasets. Our results on Librispeech are state-of- the-art compared to other hybrid approaches and competitive with previously published end-to-end numbers. Further analysis shows that chenones can better utilize powerful acoustic models and large training data, and require context-and position-dependent modeling to work well. Chenone-based systems also outperform senone baselines on proper noun and rare word recognition, an area where the latter is traditionally thought to have an advantage. Our work provides an alternative for end-to-end ASR and establishes that hybrid systems can be improved by dropping the reliance on phonetic knowledge. Index T erms-- graphemic lexicon, hybrid speech recognition, chenones, acoustic modeling, librispeech 1. INTRODUCTION In the past decade, neural network acoustic models have become a staple in automatic speech recognition (ASR).

We explore a novel setting of the Multi-Armed Bandit (MAB) problem inspired from real world applications which we call bandits with "stochastic delayed composite anonymous feedback (SDCAF)". In SDCAF, the rewards on pulling arms are stochastic with respect to time but spread over a fixed number of time steps in the future after pulling the arm. The complexity of this problem stems from the anonymous feedback to the player and the stochastic generation of the reward. Due to the aggregated nature of the rewards, the player is unable to associate the reward to a particular time step from the past. We present two algorithms for this more complicated setting of SDCAF using phase based extensions of the UCB algorithm. We perform regret analysis to show sub-linear theoretical guarantees on both the algorithms.