Recent developments in image classification and natural language processing, coupled with the rapid growth in social media usage, have enabled fundamental advances in detecting breaking events around the world in real-time. Emergency response is one such area that stands to gain from these advances. By processing billions of texts and images a minute, events can be automatically detected to enable emergency response workers to better assess rapidly evolving situations and deploy resources accordingly. To date, most event detection techniques in this area have focused on image-only or text-only approaches, limiting detection performance and impacting the quality of information delivered to crisis response teams. In this paper, we present a new multimodal fusion method that leverages both images and texts as input. In particular, we introduce a cross-attention module that can filter uninformative and misleading components from weak modalities on a sample by sample basis. In addition, we employ a multimodal graph-based approach to stochastically transition between embeddings of different multimodal pairs during training to better regularize the learning process as well as dealing with limited training data by constructing new matched pairs from different samples. We show that our method outperforms the unimodal approaches and strong multimodal baselines by a large margin on three crisis-related tasks.

Artificial intelligence technologies such as machine learning, natural-language processing, computer vision, and deep learning can help track and identify the novel coronavirus. On Dec. 31, BlueDot, a Toronto-based company that uses artificial intelligence to track the spread of infectious diseases, alerted its customers about a cluster of unusual pneumonia cases in Wuhan, China. Nine days later, the World Health Organization confirmed the discovery of a novel coronavirus, later named COVID-19, in Wuhan. Today, COVID-19 is a pandemic that has spread to 180 countries, claimed more than 50,000 lives, and triggered a near-global lockdown. And for the moment, the best solution to contain the spread of the virus is to improve personal hygiene and exercise social distancing.

Google Colab is an amazing gift to the data science community from the fine folks at Google. Colab gives us the ability to build complex and heavy machine learning and deep learning models without having to expend our machine's limited resources. I can certainly appreciate this given how much I used to struggle on my machine! The'out of memory' error is now quite infamous in the data science community – Google Colab provides us with the workaround and adds several cherries on top! I love the free GPU and TPU support – it's simply unparalleled and unrivaled in any other coding IDE.

Deep learning, a subset of machine learning and artificial intelligence (AI), has been there since a while, but became an overnight "sensation" when in 2016, Google's AI program, a robot player beat human grandmaster Lee Seedol in the famed game of AlphaGo . Since then, deep learning training and learning methods became widely acknowledged for "humanizing" machines. Many of the advanced automation capabilities now found in enterprise AI platforms are due to the rapid growth of ML and deep learning technologies, as researchers predict deep learning to provide formidable momentum for the adoption and growth of AI, even though most of these experiments are in their infancy. By definition, deep learning is a powerful tool for enterprises looking to gain actionable insights and enable automated responses to a flood of data, especially unstructured data, from all kinds of devices, Internet of Things (IoT), social media and – of course – from corporate data systems. From that perspective deep learning works incredibly well with unstructured data, such as images, sound, time-series of events and so on.

Monitoring structural damage is extremely important for sustaining and preserving the service life of civil structures. While successful monitoring provides resolute and staunch information on the health, serviceability, integrity and safety of structures; maintaining continuous performance of a structure depends highly on monitoring the occurrence, formation and propagation of damage. Damage may accumulate on structures due to different environmental and human-induced factors. Numerous monitoring and detection approaches have been developed to provide practical means for early warning against structural damage or any type of anomaly. Considerable effort has been put into vibration-based methods, which utilize the vibration response of the monitored structure to assess its condition and identify structural damage. Meanwhile, with emerging computing power and sensing technology in the last decade, Machine Learning (ML) and especially Deep Learning (DL) algorithms have become more feasible and extensively used in vibration-based structural damage detection with elegant performance and often with rigorous accuracy. While there have been multiple review studies published on vibration-based structural damage detection, there has not been a study where the transition from traditional methods to ML and DL methods are described and discussed. This paper aims to fulfill this gap by presenting the highlights of the traditional methods and provide a comprehensive review of the most recent applications of ML and DL algorithms utilized for vibration-based structural damage detection in civil structures.

We develop a method to learn physical systems from data that employs feedforward neural networks and whose predictions comply with the first and second principles of thermodynamics. The method employs a minimum amount of data by enforcing the metriplectic structure of dissipative Hamiltonian systems in the form of the so-called General Equation for the Non-Equilibrium Reversible-Irreversible Coupling, GENERIC [M. Grmela and H.C Oettinger (1997). Dynamics and thermodynamics of complex fluids. I. Development of a general formalism. Phys. Rev. E. 56 (6): 6620-6632]. The method does not need to enforce any kind of balance equation, and thus no previous knowledge on the nature of the system is needed. Conservation of energy and dissipation of entropy in the prediction of previously unseen situations arise as a natural by-product of the structure of the method. Examples of the performance of the method are shown that include conservative as well as dissipative systems, discrete as well as continuous ones.

Industrial forecasting has entered an era of unprecedented growth in the size and complexity of data which require new modeling methodologies. While many new general purpose machine learning approaches have emerged, they remain poorly understand and irreconcilable with more traditional statistical modeling approaches. We present a general class of exponential smoothed recurrent neural networks (RNNs) which are well suited to modeling non-stationary dynamical systems arising in industrial applications such as electricity load management and financial risk and trading. In particular, we analyze their capacity to characterize the non-linear partial autocorrelation structure of time series and directly capture dynamic effects such as seasonality and regime changes. Application of exponentially smoothed RNNs to electricity load forecasting, weather data and financial time series, such as minute level Bitcoin prices and CME futures tick data, highlight the efficacy of exponential smoothing for multi-step time series forecasting. The results also suggest that popular, but more complicated neural network architectures originally designed for speech processing, such as LSTMs and GRUs, are likely over-engineered for industrial forecasting and light-weight exponentially smoothed architectures capture the salient features while being superior and more robust than simple RNNs.

The inductive bias of a neural network is largely determined by the architecture and the training algorithm. To achieve good generalization, how to effectively train a neural network is even more important than designing the architecture. We propose a novel orthogonal over-parameterized training (OPT) framework that can provably minimize the hyperspherical energy which characterizes the diversity of neurons on a hypersphere. By constantly maintaining the minimum hyperspherical energy during training, OPT can greatly improve the network generalization. Specifically, OPT fixes the randomly initialized weights of the neurons and learns an orthogonal transformation that applies to these neurons. We propose multiple ways to learn such an orthogonal transformation, including unrolling orthogonalization algorithms, applying orthogonal parameterization, and designing orthogonality-preserving gradient update. Interestingly, OPT reveals that learning a proper coordinate system for neurons is crucial to generalization and may be more important than learning a specific relative position of neurons. We further provide theoretical insights of why OPT yields better generalization. Extensive experiments validate the superiority of OPT.

With the growth of interest in the attack and defense of deep neural networks, researchers are focusing more on the robustness of applying them to devices with limited memory. Thus, unlike adversarial training, which only considers the balance between accuracy and robustness, we come to a more meaningful and critical issue, i.e., the balance among accuracy, efficiency and robustness (AER). Recently, some related works focused on this issue, but with different observations, and the relations among AER remain unclear. This paper first investigates the robustness of pruned models with different compression ratios under the gradual pruning process and concludes that the robustness of the pruned model drastically varies with different pruning processes, especially in response to attacks with large strength. Second, we test the performance of mixing the clean data and adversarial examples (generated with a prescribed uniform budget) into the gradual pruning process, called adversarial pruning, and find the following: the pruned model's robustness exhibits high sensitivity to the budget. Furthermore, to better balance the AER, we propose an approach called blind adversarial pruning (BAP), which introduces the idea of blind adversarial training into the gradual pruning process. The main idea is to use a cutoff-scale strategy to adaptively estimate a nonuniform budget to modify the AEs used during pruning, thus ensuring that the strengths of AEs are dynamically located within a reasonable range at each pruning step and ultimately improving the overall AER of the pruned model. The experimental results obtained using BAP for pruning classification models based on several benchmarks demonstrate the competitive performance of this method: the robustness of the model pruned by BAP is more stable among varying pruning processes, and BAP exhibits better overall AER than adversarial pruning.

New results suggest strong limits to the feasibility of classifying human brain activity evoked from image stimuli, as measured through EEG. Considerable prior work suffers from a confound between the stimulus class and the time since the start of the experiment. A prior attempt to avoid this confound using randomized trials was unable to achieve results above chance in a statistically significant fashion when the data sets were of the same size as the original experiments. Here, we again attempt to replicate these experiments with randomized trials on a far larger (20x) dataset of 1,000 stimulus presentations of each of forty classes, all from a single subject. To our knowledge, this is the largest such EEG data collection effort from a single subject and is at the bounds of feasibility. We obtain classification accuracy that is marginally above chance and above chance in a statistically significant fashion, and further assess how accuracy depends on the classifier used, the amount of training data used, and the number of classes. Reaching the limits of data collection without substantial improvement in classification accuracy suggests limits to the feasibility of this enterprise.