Like yesterday, we bring you the best tweets covering major talks and events at IJCAI 2019. Follow the invited talk by @MichelaMilano1 at @IJCAIconf "Empirical Model Learning: merging knowledge-based and data-driven decision models through machine learning" https://t.co/FA7gR0105H Interesting idea to use deep forest ensembles as alternative to deep neural networks. Introducing the #AIglassbox: @RecklessCoding presents our paper at #ijcai2019. This takes place in 10 minutes!

As a child of refugees, my parents' narrative is missing huge gaps of information. In our data rich world, archivists are finally piecing together new clues of history using unmanned systems to reopen cold cases. The Nazis were masters in using technology to mechanize killing and erasing all evidence of their crime. Nowhere is this more apparent than in Treblinka, Poland. The death camp exterminated close to 900,000 Jews over a 15-month period before a revolt led to its dismantlement in 1943.

Bingeing your favourite boxset will be more enjoyable if you're watching it on a modern TV screen, new research suggests. Identical twin brothers were monitored by AI as they sat down to enjoy the same episode of Game of Thrones in separate rooms on different TV sets. Experimenters found that the sibling watching on the most up to date screen displayed the greatest physical and emotional responses. Bingeing your favourite boxset will be more enjoyable if you're watching it on a modern TV screen, new research suggests. Realeyes' AI platform analysed the facial expressions, head movements and body language from more than 144,000 frames of video footage captured of each twin.

Women are not inherently better at multi-tasking - and that's according to scientists. A study examining the long-asserted myth has proved that men are just as capable of juggling numerous jobs simultaneously. In fact, despite years of claims to the contrary, it transpires that both genders are equally able, or unable, to do more than one task concurrently. A team of researchers led by Dr Patricia Hirsch of Germany's Aachen University reached the conclusion after analysing 48 men and 48 women, with an average age of 24, in letter or number identification tasks. Some participants were asked to pay attention to two tasks at once, known as concurrent multitasking.

Amazon says its increasingly popular facial recognition software has learned a few new tricks, including the ability to discern when someone is scared. The software, called'Rekognition', has added'fear' to its list of detectable emotions which already includes'Happy', 'Sad', 'Angry', 'Surprised', 'Disgusted', 'Calm' and'Confused,' said Amazon in an announcement earlier this week. In addition to its emotion capabilities, Amazon says its has also improved Rekognition's ability to identify gender and age more accurately. Amazon's facial recognition software can now detect'fear' and better glean age and gender according to an announcement by the company. The improved age features offer smaller age ranges across the spectrum and also more accurate range predictions, said the company.

You could be forgiven for thinking that AI will soon replace human physicians based on headlines such as "The AI Doctor Will See You Now," "Your Future Doctor May Not Be Human," and "This AI Just Beat Human Doctors on a Clinical Exam." But experts say the reality is more of a collaboration than an ousting: Patients could soon find their lives partly in the hands of AI services working alongside human clinicians. There is no shortage of optimism about AI in the medical community. But many also caution the hype surrounding AI has yet to be realized in real clinical settings. There are also different visions for how AI services could make the biggest impact.

Model-based Reinforcement Learning (MBRL) allows data-efficient learning which is required in real world applications such as robotics. However, despite the impressive data-efficiency, MBRL does not achieve the final performance of state-of-the-art Model-free Reinforcement Learning (MFRL) methods. We leverage the strengths of both realms and propose an approach that obtains high performance with a small amount of data. In particular, we combine MFRL and Model Predictive Control (MPC). While MFRL's strength in exploration allows us to train a better forward dynamics model for MPC, MPC improves the performance of the MFRL policy by sampling-based planning. The experimental results in standard continuous control benchmarks show that our approach can achieve MFRL`s level of performance while being as data-efficient as MBRL.

Bandit algorithms have various application in safety-critical systems, where it is important to respect the system constraints that rely on the bandit's unknown parameters at every round. In this paper, we formulate a linear stochastic multi-armed bandit problem with safety constraints that depend (linearly) on an unknown parameter vector. As such, the learner is unable to identify all safe actions and must act conservatively in ensuring that her actions satisfy the safety constraint at all rounds (at least with high probability). For these bandits, we propose a new UCB-based algorithm called Safe-LUCB, which includes necessary modifications to respect safety constraints. The algorithm has two phases. During the pure exploration phase the learner chooses her actions at random from a restricted set of safe actions with the goal of learning a good approximation of the entire unknown safe set. Once this goal is achieved, the algorithm begins a safe exploration-exploitation phase where the learner gradually expands their estimate of the set of safe actions while controlling the growth of regret. We provide a general regret bound for the algorithm, as well as a problem dependent bound that is connected to the location of the optimal action within the safe set. We then propose a modified heuristic that exploits our problem dependent analysis to improve the regret.

A deep-learning-based surrogate model is developed and applied for predicting dynamic subsurface flow in channelized geological models. The surrogate model is based on deep convolutional and recurrent neural network architectures, specifically a residual U-Net and a convolutional long short term memory recurrent network. Training samples entail global pressure and saturation maps, at a series of time steps, generated by simulating oil-water flow in many (1500 in our case) realizations of a 2D channelized system. After training, the `recurrent R-U-Net' surrogate model is shown to be capable of accurately predicting dynamic pressure and saturation maps and well rates (e.g., time-varying oil and water rates at production wells) for new geological realizations. Assessments demonstrating high surrogate-model accuracy are presented for an individual geological realization and for an ensemble of 500 test geomodels. The surrogate model is then used for the challenging problem of data assimilation (history matching) in a channelized system. For this study, posterior reservoir models are generated using the randomized maximum likelihood method, with the permeability field represented using the recently developed CNN-PCA parameterization. The flow responses required during the data assimilation procedure are provided by the recurrent R-U-Net. The overall approach is shown to lead to substantial reduction in prediction uncertainty. High-fidelity numerical simulation results for the posterior geomodels (generated by the surrogate-based data assimilation procedure) are shown to be in essential agreement with the recurrent R-U-Net predictions. The accuracy and dramatic speedup provided by the surrogate model suggest that it may eventually enable the application of more formal posterior sampling methods in realistic problems.

Traditional anatomical analyses captured only a fraction of real phenomic information. Here, we apply deep learning to quantify total phenotypic similarity across 2468 butterfly photographs, covering 38 subspecies from the polymorphic mimicry complex of $\textit{Heliconius erato}$ and $\textit{Heliconius melpomene}$. Euclidean phenotypic distances, calculated using a deep convolutional triplet network, demonstrate significant convergence between interspecies co-mimics. This quantitatively validates a key prediction of M\"ullerian mimicry theory, evolutionary biology's oldest mathematical model. Phenotypic neighbor-joining trees are significantly correlated with wing pattern gene phylogenies, demonstrating objective, phylogenetically informative phenome capture. Comparative analyses indicate frequency-dependent, mutual convergence with coevolutionary exchange of wing pattern features. Therefore, phenotypic analysis supports reciprocal coevolution, predicted by classical mimicry theory but since disputed, and reveals mutual convergence as an intrinsic generator for the surprising diversity of M\"ullerian mimicry. This demonstrates that deep learning can generate phenomic spatial embeddings which enable quantitative tests of evolutionary hypotheses previously only testable subjectively.