Deep Neural Networks (DNNs) have achieved tremendous success in various tasks. However, DNNs exhibit uncertainty and unreliability when faced with well-designed adversarial examples, leading to misclassification. To address this, a variety of methods have been proposed to improve the robustness of DNNs by detecting adversarial attacks. In this paper, we combine model explanation techniques with adversarial models to enhance adversarial detection in real-world scenarios. Specifically, we develop a novel adversary-resistant detection framework called EXPLAINER, which utilizes explanation results extracted from explainable learning models. The explanation model in EXPLAINER generates an explanation map that identifies the relevance of input variables to the model’s classification result. Consequently, adversarial examples can be effectively detected by comparing the explanation results of a given sample with its denoised version, without relying on any prior knowledge of attacks. The proposed framework is thoroughly evaluated against different adversarial attacks, and experimental results demonstrate that our approach achieves promising results in white-box attack scenarios.

Question answering (QA) is a task in the field of natural language processing (NLP) and information retrieval, which has pivotal applications in areas such as online reading comprehension and web search engines. Currently, Bidirectional Encoder Representations from Transformers (BERT) and its biomedical variation (BioBERT) achieve impressive results on the reading comprehension QA datasets and medical-related QA datasets, and so they are widely used for a variety of passage-based QA tasks. However, their performances rapidly deteriorate when encountering passage and context ambiguities. This issue is prevalent and unavoidable in many fields, notably the web-based medical field. In this paper, we introduced a novel approach called the Multiple Synonymous Questions BioBERT (MSQ-BioBERT), which integrates question augmentation, rather than the typical single question used by traditional BioBERT, to elevate BioBERT’s performance on medical QA tasks. In addition, we constructed an ambiguous medical dataset based on the information from Wikipedia web. Experiments with both this web-based constructed medical dataset and open biomedical datasets demonstrate the significant performance gains of the MSQ-BioBERT approach, showcasing a new method for addressing ambiguity in medical QA tasks.

Deep learning models have achieved exceptional predictive performance in a wide variety of tasks, ranging from computer vision, natural language processing, to graph mining. Many businesses and organizations across diverse domains are now building large-scale applications based on deep learning. However, there are growing concerns, regarding the fairness, security, and trustworthiness of these models, largely due to the opaque nature of their decision processes. Recently, there has been an increasing interest in explainable deep learning that aims to reduce the opacity of a model by explaining its behavior, its predictions, or both, thus building trust between human and complex deep learning models. A collection of explanation methods have been proposed in recent years that address the problem of low explainability and opaqueness of models. In this tutorial, we introduce recent explanation methods from a data perspective, targeting models that process image data, text data, and graph data, respectively. We will compare their strengths and limitations, and offer real-world applications.

Since the first identified case of COVID-19 in December 2019, a plethora of pharmaceuticals and therapeutics have been tested for COVID-19 treatment. While medical advancements and breakthroughs are well underway, the sheer number of studies, treatments, and associated reports makes it extremely challenging to keep track of the rapidly growing COVID-19 research landscape. While existing scientific literature search systems provide basic document retrieval, they fundamentally lack the ability to explore data, and in addition, do not help develop a deeper understanding of COVID-19 related clinical experiments and findings. As research expands, results do so as well, resulting in a position that is complicated and overwhelming. To address this issue, we present a named entity recognition based framework that accurately extracts COVID-19 related information from clinical test results articles, and generates an efficient and interactive visual knowledge graph. This knowledge graph platform is user friendly, and provides intuitive and convenient tools to explore and analyze COVID-19 research data and results including medicinal performances, side effects and target populations.

Social networks can serve as a valuable communication channel for asking for help, offering assistance, and coordinating rescue activities in disaster because it allows users to continuously update critical information in the fast-changing disaster environment. This paper presents a novel sequence to sequence based framework for forecasting people’s needs during disasters using social media and weather data. It consists of two Long Short-Term Memory (LSTM) models, one of which encodes input sequences of weather information and the other plays as a conditional decoder that decodes the encoded vector and forecasts the survivors’ needs. Case studies using data collected during Hurricane Sandy in 2012, Hurricane Harvey and Hurricane Irma in 2017 demonstrate that the proposed approach outperformed the statistical language model n-gram, LSTM generative model, and convolutional neural network (CNN) based model. This research indicates its great promise for enhancing disaster management such as evacuation planning and commodity delivery.

Despite the popularity of Convolutional Neural Networks (CNN), the problem of uncertainty quantification (UQ) of CNN has been largely overlooked. Lack of efficient UQ tools severely limits the application of CNN in certain areas, such as medicine, where prediction uncertainty is critically important. Among the few existing UQ approaches that have been proposed for deep learning, none of them has theoretical consistency that can guarantee the uncertainty quality. To address this issue, we propose a novel bootstrap based framework for the estimation of prediction uncertainty. The inference procedure we use relies on convexified neural networks to establish the theoretical consistency of bootstrap. Our approach has a significantly less computational load than its competitors, as it relies on warm-starts at each bootstrap that avoids refitting the model from scratch. We further explore a novel transfer learning method so our framework can work on arbitrary neural networks. We experimentally demonstrate our approach has a much better performance compared to other baseline CNNs and state-of-the-art methods on various image datasets.

Driver classification is used recently for vehicle anti-burglary and fake driver accounts based on driving behavior. Anti-burglary is a challenging problem as it leans on external devices to defend against vehicle theft. Several researchers analyzed the driving behavior to identify drivers, but they faced several challenges to produce a stable model for the cold start problem and for medium-long sequences. In addition, some approaches had an unpleasant performance when the action space increased (> 2 drivers). In this paper, we propose a novel approach named OnlineDC (Online Driver Classification), which leverages temporal driving behavior to identify a human subject behind the wheel. Our method utilizes the Gated Recurrent Unit (GRU) and the ResNet with the Squeeze-Excite blocks (SE) to analyze the long-short term patterns of driving behaviors. Moreover, we fostered the performance by building and applying the Feature Generation (FG) algorithm to extract spectral, temporal, and statistical features from the sensing data of vehicles. We conducted extensive experiments to show how our approach outperformed state-of-the-art baseline methods. The results also showed that our solution could resolve the cold-start problem for short patterns.

The growing use of deep learning for a wide range of data problems has highlighted the need to understand and diagnose these models appropriately, making deep learning interpretation techniques an essential tool for data analysts. The numerous model interpretation methods proposed in recent years are generally based on heuristics, with little or no theoretical guarantees. Here, we present a statistical framework for saliency estimation for black-box computer vision models. Our proposed model-agnostic estimation procedure, which is statistically consistent and capable of passing sanity checks, has polynomial-time computational efficiency since it only requires solving a linear program. An upper bound is established on the number of model evaluations needed to recover regions of importance with high probability through our theoretical analysis. Furthermore, a new perturbation scheme is presented for the estimation of local gradients that is more efficient than commonly used random perturbation schemes. The validity and excellence of our new method are demonstrated experimentally via sensitivity analyses.

This project built a world-wide database of coron-avirus cases, which helps to model the spread of the coronavirus disease (COVID-19), and to identify policy and social factors that impact the spread of COVID-19. Four essential tasks are implemented: 1) build a comprehensive database of coronavirus cases world-wide; 2) visualize the heatmap of confirmed cases for each country, provide detailed spreading trends for each countries and comparison among countries; 3) collect tweets about COVID-19 in real-time and extract people's daily concern flow; 4) integrate breaking news such as first confirmed/death case in each country. This demo will provide decision-makers with accurate data-driven representations in an easy to understand format that enables them to make more timely and cost-effective preparation and response plans.

Chronic alcohol consumption has become one of the major concerns of the society in the sector of public health. The death toll due to excessive consumption of alcohol is increasing exponentially. Face-to-face interaction to create awareness and to stymie the consumption of alcohol is a quixotic solution. Alcohol relapse remains a challenging problem in disorders associated with alcohol addiction which is related to spatial-temporal factors like periods and specific places. In this paper, a new generation of relapse prevention mobile application called “BeSober” is proposed. It assists users to develop abstemious habits in drinking, providing support within the intervention period, tracking risky alcoholic spots, presenting community-based support from alternative users or alcohol addiction therapists, monitoring addict's behaviour and offer personalized recommendations to assist the addict in staying sober.

Ahstract-A crucial and time-sensitive task when any disaster occurs is to rescue victims and distribute resources to the right groups and locations. This task is challenging in populated urban areas, due to a huge burst of help requests made in a very short period. To improve the efficiency of the emergency response in the immediate aftermath of a disaster, we propose a heuristic multi-agent reinforcement learning scheduling algorithm, named as ResQ, which can effectively schedule a rapid deployment of volunteers to rescue victims in dynamic settings. The core concept is to quickly identify victims and volunteers from social network data and then schedule rescue parties with an adaptive learning algorithm. This framework performs two key functions: 1) identify trapped victims and volunteers, and 2) optimize the volunteers' rescue strategy in a complex time-sensitive environment. The proposed ResQ algorithm can speed up the training processes through a heuristic function which reduces the state-action space by identifying a set of particular actions over others. Experimental results showed that the proposed heuristic multi-agent reinforcement learning based scheduling outperforms several state-of-art methods, in terms of both reward rate and response times.

The Coronavirus Disease 2019 (COVID-19) has now become a pandemic, inflicting millions of people and causing tens of thousands of deaths. To better understand the dynamics of COVID-19, we present a comprehensive COVID-19 tracking and visualization platform that pinpoints the dynamics of the COVID-19 worldwide. Four essential components are implemented: 1) presenting the visualization map of COVID-19 confirmed cases and total counts all over the world; 2) showing the worldwide trends of COVID-19 at multi-grained levels; 3) provide multiview comparisons, including confirmed cases per million people, mortality rate and accumulative cure rate; 4) integrating a multi-grained view of the disease spreading dynamics in China and showing how the epidemic is taken under control in China. This demo will spur further disease spreading modeling for researchers, support decision-maker, and enrich the public awareness of the spreading situations of COVID-19 worldwide. This demo offers decision-makers with accurate data-driven representations in an easy to understand format that informs them the pandemic dynamics and enables them to make more timely and cost-effective preparation and response plans.

Automating pneumonia diagnosis from X-ray images could significantly improve patient diagnosing outcomes. A major challenge is that disease information (features) must be extracted directly from the image backgrounds. Motivated by recent advances in Convolutional Neural Network (CNN), we propose a hierarchical weighting deep learning model, ChestWNet, that combines DenseNet and transfer learning to detect and localize thoracic diseases from chest x-rays. Hierarchical weighting networks are designed to assign scores reflecting the importance of specific pixels (regions), and learning weights at pixel-, region-, and image-levels, jointly learning these hierarchical weighting networks and the image classification network in an end-to-end manner. Chest X-ray datasets are customized to solve the unbalancing label problem in these datasets. Extensive experiments show that ChestWNet significantly outperforms other established prediction methods, and can also be applied to similar scenarios with fixed point-of-interest regions in images.

Anti-theft problem has been challenging since it mainly depends on the existence of external devices to defend from thefts. Recently, driver behavior analysis using supervised learning has been investigated with the goal to detect burglary by identifying drivers. In this paper, we propose a data-driven technique, LiveDI, which uses driving behavior removing the use of external devices in order to identify drivers. The built model utilizes Gated Recurrent Unit (GRU) and Fully Convolutional Networks (FCN) to learn long-short term patterns of the driving behaviors from drivers. Additionally, we improve the training time by utilizing the Segmented Feature Generation (SFG) algorithm to reduce the state space where the driving behaviors are split with a time window for analysis. Extensive experiments are conducted which show the impact of parameters on our technique and verify that our proposed approach outperforms the state-of-the-art baseline methods.

Opioid addiction is a severe public health threat in the U.S, causing massive deaths and many social problems. Accurate relapse prediction is of practical importance for recovering patients since relapse prediction promotes timely relapse preventions that help patients stay clean. In this paper, we introduce a Generative Adversarial Networks (GAN) model to predict the addiction relapses based on sentiment images and social influences. Experimental results on real social media data from Reddit.com demonstrate that the GAN model delivers a better performance than comparable alternative techniques. The sentiment images generated by the model show that relapse is closely connected with two emotions 'joy' and 'negative'. This work is one of the first attempts to predict relapses using massive social media (Reddit.com) data and generative adversarial nets. The proposed method, combined with knowledge of social media mining, has the potential to revolutionize the practice of opioid addiction prevention and treatment.

During disasters, discover people's concerns dynamically is crucial to disaster rescue and relief. In this paper, we propose a social media based framework to analyze people's concerns, to access the importance and to track the dynamic changes of these concerns. To better understand people's concerns across platforms and to monitor the dynamics, we make comparisons between Tweets and news on the mentioned aspects and disclosed some interesting findings. Specifically, we take 2018 Camp Fire, the most destructive wildfire on record in history of California as a case study. We find that despite their keen attentions towards the disaster, social media and news media focus on different aspects of the disaster, so are the contents and dynamic changes of their concerns.

It is of paramount importance to know the situations of people who undergone disaster events and be aware of their updates, yet it is not an easy job to accomplish in the chaos of a disaster. To facilitate advanced disaster relief organization and efficient supplement distribution, we develop NiPred, a social media based need prediction prototype that predicts needs for victims across the affected area. NiPred first extracts problems and concerns posted by victims of disaster-hurricane, in our case study; then displays the statistics to offer an overview for awareness and further analysis; and last, predicts the needs such as "diaper", "boat", "canoe" and "shelter" etc. for disaster relief planning.

It is widely acknowledged that addiction relapse is highly associated with spatial-temporal factors such as some specific places or time periods. Current studies suggest that those factors can be utilized for better relapse interventions, however, there is no relapse prevention application that makes use of those factors. In this paper, we introduce a mobile app called "Addict Free", which records user profiles, tracks relapse history and summarizes recovering statistics to help users better understand their recovering situations. Also, this app builds a relapse recovering community, which allows users to ask for advice and encouragement, and share relapse prevention experience. Moreover, machine learning algorithms that ingest spatial and temporal factors are utilized to predict relapse, based on which helpful addiction diversion activities are recommended by a recovering recommendation algorithm. By interacting with users, this app targets at providing smart suggestions that aim to stop relapse, especially for alcohol and tobacco addiction users.

Water is a critical natural resource that has significant impacts on human living and society. Growing population and energy consumption exacerbate the scarcity of water and our ability to manage this resource. This demonstration paper presents WaterScope, a smart and connected platform for water resource management, which integrates multiple data sources such as water level data, social media data, and water related articles. Furthermore, the tool enables forecasting underground water levels, identifying water concerns, sharing knowledge and expertise among stakeholders, and thus bringing new insights to our understanding and insights of the water supplies and resource management. The prototype engages water stakeholders who face problems of similar nature but deal with the problem in an ad-hoc and isolated manner. The interactive WaterScope platform targets creating an interconnected virtual community that aims to improve water supply resilience.

It is imperatively important to spot hidden road hazards which cause a high proportion of traffic incidents. Social media data is featured by its innumerable and up-to-date information, and provides a promising approach to road hazard spotting. However, this research area is not well studied yet. In this paper, we present our view of the important research issues, including challenges of mining spatio-temporal dataset, road hazards reasoning and etc.

Precisely forecasting wind speed is essential for wind power producers and grid operators. However, this task is challenging due to the stochasticity of wind speed. To accurately predict short-term wind speed under uncertainties, this paper proposed a multi-variable stacked LSTMs model (MSLSTM). The proposed method utilizes multiple historical meteorological variables, such as wind speed, temperature, humidity, pressure, dew point and solar radiation to accurately predict wind speeds. The prediction performance is extensively assessed using real data collected in West Texas, USA. The experimental results show that the proposed MSLSTM can preferably capture and learn uncertainties while output competitive performance.

Social media has become a popular data source to track and analyze societal events. Targeted domains such as election, civil unrest, and spreading disease all require a natural language normalization tool capable of extracting information pertinent to these domains accurately. Due to the unstructured language, short-length messages, casual posting styles, and homonyms, it is technically difficult and labor-intensive to remove barriers that may lead to inaccurate analysis. Because the fact that typos or other symbolic representations of sentiment may lead to lower frequency of term appearance, language preprocessing becomes critical and necessary to improve social media text reasoning. We propose a novel unsupervised preprocessing approach to enhance text understanding quality and illustrate this approach using one specific domain, flu shot reasoning. The proposed approach relies on a database of synonyms and opposite words and an algorithm to transform negative sentences into its affirmative form. In this form, the features and opinions are reflected accurately via transforming parts of speech. For instance, features are presented as nouns and opinions are presented as verbs or adjectives. The algorithm also corrects words if they are not correctly written and normalizes them to increase its frequency of appearance. The effectiveness of our algorithm is evaluated on the tweets dataset to answer why people are reluctant to take flu shots.

Social media provided a significant communication platform for rescuing people when Harvey hit Houston area. In this devastating flooding crisis, the overloaded official emergency institutes were not able to respond quickly due to the burst of call for help in a very short period of time. In this circumstance, many volunteers and people who needed help often post their information on social media such as Twitter and Facebook. How to organize volunteers smartly and efficiently to help people is an extremely challenging and significant problem considering the constraints of volunteer's time slots, urgent priorities, etc. In this paper, we propose three rescue scheduling algorithms to explain how to provide victims timely help by the volunteers on social media.

Event detection in online social media has primarily focused on identifying abnormal spikes, or bursts, in activity. However, disruptive events such as socio-economic disasters, civil unrest, and even power outages, often involve abnormal troughs or lack of activity, leading to absenteeism. We present the first study, to our knowledge, that models absenteeism and uses detected absenteeism instances as a basis for event detection in location-based social networks such as Twitter. The proposed framework addresses the challenges of (i) early detection of absenteeism, (ii) identifying the locus of the absenteeism, and (iii) identifying groups or communities underlying the absenteeism. Our approach uses the formalism of graph wavelets to represent the spatiotemporal structure of user activity in a location-based social network. This formalism facilitates multiscale analysis, enabling us to detect anomalous behavior at different graph resolutions, which in turn allows the identification of event locations and underlying groups. The effectiveness of our approach is evaluated using Twitter activity related to civil unrest events in Latin America.

Modeling the movement of information within social media outlets, like Twitter, is key to understanding to how ideas spread but quantifying such movement runs into several difficulties. Two specific areas that elude a clear characterization are (i) the intrinsic random nature of individuals to potentially adopt and subsequently broadcast a Twitter topic, and (ii) the dissemination of information via non-Twitter sources, such as news outlets and word of mouth, and its impact on Twitter propagation. These distinct yet inter-connected areas must be incorporated to generate a comprehensive model of information diffusion. We propose a bispace model to capture propagation in the union of (exclusively) Twitter and non-Twitter environments. To quantify the stochastic nature of Twitter topic propagation, we combine principles of geometric Brownian motion and traditional network graph theory. We apply Poisson process functions to model information diffusion outside of the Twitter mentions network. We discuss techniques to unify the two sub-models to accurately model information dissemination. We demonstrate the novel application of these techniques on real Twitter datasets related to mass protest adoption in social communities.

Financial markets are quite sensitive to unanticipated news and events. Identifying the effect of news on the market is a challenging task. In this demo, we present Forex-foreteller (FF) which mines news articles and makes forecasts about the movement of foreign currency markets. The system uses a combination of language models, topic clustering, and sentiment analysis to identify relevant news articles. These articles along with the historical stock index and currency exchange values are used in a linear regression model to make forecasts. The system has an interactive visualizer designed specifically for touch-sensitive devices which depicts forecasts along with the chronological news events and financial data used for making the forecasts.