Few-shot molecular property prediction is a persisting challenge in many biology-related tasks, because the same molecule may exhibit different properties (e.g., active or inactive) in different tasks. Existing methods view all atoms as equally important and attend to predict the properties by averaging the features of similar molecules, which ignores key substructures within molecules and leads to poor prediction performance. Since a molecule implicitly includes key substructures, which determines the properties of the molecular, and the atom combination forms key substructures, we focus on the atom combination in this paper. With this, we propose the Hierarchical Molecular Attention Network (HMAN) to predict the molecular properties through combining atoms. First, we utilize the average pooling to extract both the prototype and the molecular features, and use Graph Neural Networks (GNN) to extract the atomic features, then concatenates the prototype and molecular features with all atomic features as input to the self-attention mechanism to calculate the different weights for atoms. Here, we select the top $B$ atoms with the highest attention scores to form the key substructures. Second, we view the formed key substructures as the query vectors and regard the molecular features as the key-value pairs, and then feed them into another self-attention to obtain the scores of key substructures. We still select $k$ key substructures with the highest scores, and weight the sum of a molecular feature and $k$ key substructures to predict the molecular properties. To train the HMAN, we design a new loss function, which includes Binary Cross-Entropy and a weighted negative log-likelihood. The former is to predict the molecular properties, and the last one is to optimize the weight distribution, rendering that the weight distribution matches the predicted probability distribution, with back-propagation. Theoretical analysis proves the convergence of the new loss function, and extensive experimental results demonstrate that HMAN significantly outperforms SOTA baseline models in molecular property prediction tasks.

We study the problem of classifying different cooking styles, based on the recipe. The difficulty is that the same food ingredients, seasoning, and the very similar instructions result in different flavors, with different cooking styles. Existing methods have limitations: they mainly focus on homogeneous data (e.g., instruction or image), ignoring heterogeneous data (e.g., flavor compound or ingredient), which certainly hurts the classification performance. This is because collecting enough available heterogeneous data of a recipe is a non-trivial task. In this paper, we present a new heterogeneous data augmentation method to improve classification performance. Specifically, we first construct a heterogeneous recipe graph network to represent heterogeneous data, which includes four main-stream types of heterogeneous data: ingredient, flavor compound, image, and instruction. Then, we draw a sequence of augmented graphs for Semi-Supervised learning through multinomial sampling. The probability distribution of sampling depends on the Cosine distance between the nodes of graph. In this way, we name our approach as Multinomial Sampling Graph Data Augmentation (MS-GDA). Extensive experiments demonstrate that MS-GDA significantly outperforms SOTA baselines on cuisine classification and region prediction with the recipe benchmark dataset. Code is available at https://github.com/LiangzheChen/MS-GDA .

Recent hurricane events have caused unprecedented amounts of damage on critical infrastructure systems and have severely threatened our public safety and economic health. The most observable (and severe) impact of these hurricanes is the loss of electric power in many regions, which causes breakdowns in essential public services. Understanding power outages and how they evolve during a hurricane provides insights on how to reduce outages in the future, and how to improve the robustness of the underlying critical infrastructure systems. In this article, we propose a novel scalable segmentation with explanations framework to help experts understand such datasets. Our method, CnR (Cut-n-Reveal), first finds a segmentation of the outage sequences based on the temporal variations of the power outage failure process so as to capture major pattern changes. This temporal segmentation procedure is capable of accounting for both the spatial and temporal correlations of the underlying power outage process. We then propose a novel explanation optimization formulation to find an intuitive explanation of the segmentation such that the explanation highlights theculprittime series of the change in each segment. Through extensive experiments, we show that our method consistently outperforms competitors in multiple real datasets with ground truth. We further study real county-level power outage data from several recent hurricanes (Matthew, Harvey, Irma) and show that CnR recovers important, non-trivial, and actionable patterns for domain experts, whereas baselines typically do not give meaningful results.

Microblogging websites, like Twitter and Weibo, are used by billions of people to create and spread information. This activity depends on various factors such as the friendship links between users, their topic interests and social influence between them. Social influence can be thought of as a latent factor, that may alter users posting and linking behaviors. Making sense of these behaviors is very important for fully understanding and utilizing these platforms. Most prior work in this space either ignores the effect of social influence, or considers its effect only on link formation or post generation. In contrast, we propose PoLIM, leveraging simple weak supervision, a novel model which jointly models the effect of influence on both link and post generation. We also give PoLIM-FIT, an efficient parallel inference algorithm which scales to large datasets. In our experiments on a large tweets corpus, we detect meaningful topical communities, celebrities, as well as the influence strengths patterns among them. Further, we find that there are significant portions of posts and links that are caused by influence, and this portion increases when the data focuses on a specific event. We also show that differentiating and identifying these influenced content benefits other specific quantitative downstream tasks as well, like predicting future tweets and link formation, where we significantly outperform state-of-the-art.

Segmenting temporal data sequences is an important problem which helps in understanding data dynamics in multiple applications such as epidemic surveillance, motion capture sequences, etc. In this paper, we give DASSA, the first self-guided and efficient algorithm to automatically find a segmentation that best detects the change of pattern in data sequences. To avoid introducing tuning parameters, we design DASSA to be a multi-level method which examines segments at each level of granularity via a compact data structure called the segment-graph. We build this data structure by carefully leveraging the information bottleneck method with the MDL principle to effectively represent each segment.Next, DASSA efficiently finds the optimal segmentation via a novel average-longest-path optimization on the segment-graph. Finally we show how the outputs from DASSA can be naturally interpreted to reveal meaningful patterns. We ran DASSA on multiple real datasets of varying sizes and it is very effective in finding the time-cut points of the segmentations (in some cases recovering the cut points perfectly) as well as in finding the corresponding changing patterns.

Given a sequence of snapshots of flu propagating over a population network, can we find a segmentation when the patterns of the disease spread change, possibly due to interventions? In this paper, we study the problem of segmenting graph sequences with labeled nodes. Memes on the Twitter network, diseases over a contact network, movie-cascades over a social network, etc. are all graph sequences with labeled nodes. Most related work on this subject is on plain graphs and hence ignores the label dynamics. Others require fix parameters or feature engineering. We propose SNAPNETS, to automatically find segmentations of such graph sequences, with different characteristics of nodes of each label in adjacent segments. It satisfies all the desired properties (being parameter free, comprehensive and scalable) by leveraging a principled, multi-level, flexible framework which maps the problem to a path optimization problem over a weighted DAG. Also, we develop the parallel framework of SNAPNETS which speeds up its running time. Finally, we propose an extension of SNAPNETS to handle the dynamic graph structures and use it to detect anomalies (and events) in network sequences. Extensive experiments on several diverse real datasets show that it finds cut points matching ground-truth or meaningful external signals and detects anomalies outperforming non-trivial baselines. We also show that the segmentations are easily interpretable, and that SNAPNETS scales near-linearly with the size of the input. Finally, we show how to use SNAPNETS to detect anomaly in a sequence of dynamic networks.

Given a sequence of snapshots of flu propagating over a population network, can we find a segmentation when the patterns of the disease spread change, possibly due to interventions? In this paper, we study the problem of segmenting graph sequences with labeled nodes. Memes on the Twitter network, diseases over a contact network, movie-cascades over a social network, etc. are all graph sequences with labeled nodes. Most related work is on plain graphs (and hence ignore the label dynamics) or fix parameters or require much feature engineering. Instead, we propose SnapNETS, to automatically find segmentations of such graph sequences, with different characteristics of nodes of each label in adjacent segments. It satisfies all the desired properties (being parameter-free, comprehensive and scalable) by leveraging a principled, multi-level, flexible framework which maps the problem to a path optimization problem over a weighted DAG. Extensive experiments on several diverse real datasets show that it finds cut points matching ground-truth or meaningful external signals outperforming non-trivial baselines. We also show that SnapNETS scales near-linearly with the size of the input.

Critical Infrastructure Systems such as transportation, water and power grid systems are vital to our national security, economy, and public safety. Recent events, like the 2012 hurricane Sandy, show how the interdependencies among different CI networks lead to catastrophic failures among the whole system. Hence, analyzing these CI networks, and modeling failure cascades on them becomes a very important problem. However, traditional models either do not take multiple CIs or the dynamics of the system into account, or model it simplistically. In this paper, we study this problem using a heterogeneous network viewpoint. We first construct heterogeneous CI networks with multiple components using national-level datasets. Then we study novel failure maximization problems on these networks, to compute critical nodes in such systems. We then provide HotSpots, a scalable and effective algorithm for these problems, based on careful transformations. Finally, we conduct extensive experiments on real CIS data from multiple US states, and show that our method HotSpots outperforms non-trivial baselines, gives meaningful results and that our approach gives immediate benefits in providing situational-awareness during large-scale failures.

Critical Infrastructures (CIs) such as energy, water, and transportation are complex networks that are crucial for sustaining day-to-day commodity flows vital to national security, economic stability, and public safety. The nature of these CIs is such that failures caused by an extreme weather event or a man-made incident can trigger widespread cascading failures, sending ripple effects at regional or even national scales. To minimize such effects, it is critical for emergency responders to identify existing or potential vulnerabilities within CIs during such stressor events in a systematic and quantifiable manner and take appropriate mitigating actions. We present here a novel critical infrastructure monitoring and analysis system named URBAN-NET. The system includes a software stack and tools for monitoring CIs, pre-processing data, interconnecting multiple CI datasets as a heterogeneous network, identifying vulnerabilities through graph-based topological analysis, and predicting consequences based on “what-if” simulations along with visualization. As a proof-of-concept, we present several case studies to show the capabilities of our system. We also discuss remaining challenges and future work.

Surveillance of epidemic outbreaks and spread from social media is an important tool for governments and public health authorities. Machine learning techniques for now casting the flu have made significant inroads into correlating social media trends to case counts and prevalence of epidemics in a population. There is a disconnect between data-driven methods for forecasting flu incidence and epidemiological models that adopt a state based understanding of transitions, that can lead to sub-optimal predictions. Furthermore, models for epidemiological activity and social activity like on Twitter predict different shapes and have important differences. We propose a temporal topic model to capture hidden states of a user from his tweets and aggregate states in a geographical region for better estimation of trends. We show that our approach helps fill the gap between phenomenological methods for disease surveillance and epidemiological models. We validate this approach by modeling the flu using Twitter in multiple countries of South America. We demonstrate that our model can consistently outperform plain vocabulary assessment in flu case-count predictions, and at the same time get better flu-peak predictions than competitors. We also show that our fine-grained modeling can reconcile some contrasting behaviors between epidemiological and social models.