This is another Free Coursera course to learn how deep learning with neural networks can be used to classify images and detect objects in images and videos. In this course, you will use convolutional neural networks (CNNs) to classify images and detect objects.

Abstract: A common way to train neural networks is the Backpropagation. This algorithm includes a gradient descent method, which needs an adaptive step size. In the area of neural networks, the ADAM-Optimizer is one of the most popular adaptive step size methods. The 5865 citations in only three years shows additionally the importance of the given paper. We discovered that the given convergence proof of the optimizer contains some mistakes, so that the proof will be wrong.

Physics-constrained machine learning is emerging as an important topic in the field of machine learning for physics. One of the most significant advantages of incorporating physics constraints into machine learning methods is that the resulting model requires significantly less data to train. By incorporating physical rules into the machine learning formulation itself, the predictions are expected to be physically plausible. Gaussian process (GP) is perhaps one of the most common methods in machine learning for small datasets. In this paper, we investigate the possibility of constraining a GP formulation with monotonicity on three different material datasets, where one experimental and two computational datasets are used. The monotonic GP is compared against the regular GP, where a significant reduction in the posterior variance is observed. The monotonic GP is strictly monotonic in the interpolation regime, but in the extrapolation regime, the monotonic effect starts fading away as one goes beyond the training dataset. Imposing monotonicity on the GP comes at a small accuracy cost, compared to the regular GP. The monotonic GP is perhaps most useful in applications where data is scarce and noisy, and monotonicity is supported by strong physical evidence.

We present an approach for interpreting a black-box alarming system for forecasting accidents and anomalies during the drilling of oil and gas wells. The interpretation methodology aims to explain the local behavior of the accident predictive model to drilling engineers. The explanatory model uses Shapley additive explanations analysis of features, obtained through Bag-of-features representation of telemetry logs used during the drilling accident forecasting phase. Validation shows that the explanatory model has 15% precision at 70% recall, and overcomes the metric values of a random baseline and multi-head attention neural network. These results justify that the developed explanatory model is better aligned with explanations of drilling engineers, than the state-of-the-art method. The joint performance of explanatory and Bag-of-features models allows drilling engineers to understand the logic behind the system decisions at the particular moment, pay attention to highlighted telemetry regions, and correspondingly, increase the trust level in the accident forecasting alarms.

Deep neural networks (DNNs) are increasingly being applied in malware detection and their robustness has been widely debated. Traditionally an adversarial example generation scheme relies on either detailed model information (gradient-based methods) or lots of samples to train a surrogate model, neither of which are available in most scenarios. We propose the notion of the instance-based attack. Our scheme is interpretable and can work in a black-box environment. Given a specific binary example and a malware classifier, we use the data augmentation strategies to produce enough data from which we can train a simple interpretable model. We explain the detection model by displaying the weight of different parts of the specific binary. By analyzing the explanations, we found that the data subsections play an important role in Windows PE malware detection. We proposed a new function preserving transformation algorithm that can be applied to data subsections. By employing the binary-diversification techniques that we proposed, we eliminated the influence of the most weighted part to generate adversarial examples. Our algorithm can fool the DNNs in certain cases with a success rate of nearly 100\%. Our method outperforms the state-of-the-art method . The most important aspect is that our method operates in black-box settings and the results can be validated with domain knowledge. Our analysis model can assist people in improving the robustness of malware detectors.

We revisit heavy-tailed corrupted least-squares linear regression assuming to have a corrupted $n$-sized label-feature sample of at most $\epsilon n$ arbitrary outliers. We wish to estimate a $p$-dimensional parameter $b^*$ given such sample of a label-feature pair $(y,x)$ satisfying $y=\langle x,b^*\rangle+\xi$ with heavy-tailed $(x,\xi)$. We only assume $x$ is $L^4-L^2$ hypercontractive with constant $L>0$ and has covariance matrix $\Sigma$ with minimum eigenvalue $1/\mu^2>0$ and bounded condition number $\kappa>0$. The noise $\xi$ can be arbitrarily dependent on $x$ and nonsymmetric as long as $\xi x$ has finite covariance matrix $\Xi$. We propose a near-optimal computationally tractable estimator, based on the power method, assuming no knowledge on $(\Sigma,\Xi)$ nor the operator norm of $\Xi$. With probability at least $1-\delta$, our proposed estimator attains the statistical rate $\mu^2\Vert\Xi\Vert^{1/2}(\frac{p}{n}+\frac{\log(1/\delta)}{n}+\epsilon)^{1/2}$ and breakdown-point $\epsilon\lesssim\frac{1}{L^4\kappa^2}$, both optimal in the $\ell_2$-norm, assuming the near-optimal minimum sample size $L^4\kappa^2(p\log p + \log(1/\delta))\lesssim n$, up to a log factor. To the best of our knowledge, this is the first computationally tractable algorithm satisfying simultaneously all the mentioned properties. Our estimator is based on a two-stage Multiplicative Weight Update algorithm. The first stage estimates a descent direction $\hat v$ with respect to the (unknown) pre-conditioned inner product $\langle\Sigma(\cdot),\cdot\rangle$. The second stage estimate the descent direction $\Sigma\hat v$ with respect to the (known) inner product $\langle\cdot,\cdot\rangle$, without knowing nor estimating $\Sigma$.

The expected goal provides a more representative measure of the team and player performance which also suit the low-scoring nature of football instead of score in modern football. The score of a match involves randomness and often may not represent the performance of the teams and players, therefore it has been popular to use the alternative statistics in recent years such as shots on target, ball possessions, and drills. To measure the probability of a shot being a goal by the expected goal, several features are used to train an expected goal model which is based on the event and tracking football data. The selection of these features, the size and date of the data, and the model which are used as the parameters that may affect the performance of the model. Using black-box machine learning models for increasing the predictive performance of the model decreases its interpretability that causes the loss of information that can be gathered from the model. This paper proposes an accurate expected goal model trained consisting of 315,430 shots from seven seasons between 2014-15 and 2020-21 of the top-five European football leagues. Moreover, this model is explained by using explainable artificial intelligence tool to obtain an explainable expected goal model for evaluating a team or player performance. To the best of our knowledge, this is the first paper that demonstrates a practical application of an explainable artificial intelligence tool aggregated profiles to explain a group of observations on an accurate expected goal model for monitoring the team and player performance. Moreover, these methods can be generalized to other sports branches.

Estimating uncertainty is at the core of performing scientific measurements in HEP: a measurement is not useful without an estimate of its uncertainty. The goal of uncertainty quantification (UQ) is inextricably linked to the question, "how do we physically and statistically interpret these uncertainties?" The answer to this question depends not only on the computational task we aim to undertake, but also on the methods we use for that task. For artificial intelligence (AI) applications in HEP, there are several areas where interpretable methods for UQ are essential, including inference, simulation, and control/decision-making. There exist some methods for each of these areas, but they have not yet been demonstrated to be as trustworthy as more traditional approaches currently employed in physics (e.g., non-AI frequentist and Bayesian methods). Shedding light on the questions above requires additional understanding of the interplay of AI systems and uncertainty quantification. We briefly discuss the existing methods in each area and relate them to tasks across HEP. We then discuss recommendations for avenues to pursue to develop the necessary techniques for reliable widespread usage of AI with UQ over the next decade.

Customer lifetime value (CLV or LTV) refers broadly to the revenue that a company can attribute to one or more customer over the length of their relationship with the company [55]. The process of predicting the lifetime value consists in producing one or more monetary values that correspond to the sum of all the different types of revenues that a specific customer, or a specific cohort, will generate in the future. The purposes of this prediction are manifold: for example, having an early estimation of a customer's potential value allows more accurate budgeting for future investment; moreover, monitoring the remaining potential revenue from an established customer could permit preemptive actions in case of decreased engagement. Predicting customer lifetime value is a complex challenge and, to date, there is no single established practice. Furthermore, due to its wide potential impact in different business aspects, the problem is being researched in different communities using a plethora of different techniques, varying from parametric statistical models to deep learning [28, 70].

The accurate and personalized estimation of task difficulty provides many opportunities for optimizing user experience. However, user diversity makes such difficulty estimation hard, in that empirical measurements from some user sample do not necessarily generalize to others. In this paper, we contribute a new approach for personalized difficulty estimation of game levels, borrowing methods from content recommendation. Using factorization machines (FM) on a large dataset from a commercial puzzle game, we are able to predict difficulty as the number of attempts a player requires to pass future game levels, based on observed attempt counts from earlier levels and levels played by others. In addition to performance and scalability, FMs offer the benefit that the learned latent variable model can be used to study the characteristics of both players and game levels that contribute to difficulty. We compare the approach to a simple non-personalized baseline and a personalized prediction using Random Forests. Our results suggest that FMs are a promising tool enabling game designers to both optimize player experience and learn more about their players and the game.