Priyadarshini Kumari

Over the last few years Literature-based Discovery (LBD) has regained popularity as a means to enhance the scientific research process. The resurgent interest has spurred the development of supervised and semi-supervised machine learning models aimed at making previously implicit connections between scientific concepts/entities explicit based on often extensive repositories of published literature. Understanding the temporally evolving interactions between these entities can provide valuable information for predicting the future development of entity relationships. However, existing methods often underutilize the latent information embedded in the temporal aspects of interaction data. In this context, motivated by applications in the food domain—where we aim to connect nutritional information with health-related benefits—we address the hypothesis-generation problem using a temporal graph-based approach. Given that hypothesis generation involves predicting future (i.e., still to be discovered) entity connections, the ability to capture the dynamic evolution of connections over time is pivotal for a robust model. To address this, we introduce THiGER, a novel batch contrastive temporal node-pair embedding method. THiGER excels in providing a more expressive node-pair encoding by effectively harnessing node-pair relationships. Furthermore, we present THiGER-A, an incremental training approach that incorporates an active curriculum learning strategy to mitigate label bias arising from unobserved connections. By progressively training on increasingly challenging and high-utility samples, our approach significantly enhances the performance of the embedding model. Empirical validation of our proposed method demonstrates its effectiveness on established temporal-graph benchmark datasets, as well as on real-world datasets within the food domain.

Research on knowledge graph completion (KGC)—i.e., link prediction within incomplete KGs—is witnessing significant growth in popularity. Recently, KGC using KG embedding (KGE) models, primarily based on complex architectures (e.g., transformers), have achieved remarkable performance. Still, extracting the \emphminimal and relevant information employed by KGE models to make predictions, while constituting a major part of \emphexplaining the predictions, remains a challenge. While there exists a growing literature on explainers for trained KGE models, systematically exposing and quantifying their failure cases poses even greater challenges. In this work, we introduce two synthetic datasets, FRUNI and FTREE, designed to demonstrate the (in)ability of explainer methods to spot link predictions that rely on indirectly connected links. Notably, we empower practitioners to control various aspects of the datasets, such as noise levels and dataset size, enabling them to assess the performance of explainability methods across diverse scenarios. Through our experiments, we assess the performance of four recent explainers in providing accurate explanations for predictions on the proposed datasets. We believe that these datasets are valuable resources for further validating explainability methods within the knowledge graph community.

For humans and other animals, the sense of smell provides crucial information in many situations of everyday life. Still, the study of olfactory perception has received only limited attention outside of the biological sciences. From an AI perspective, the complexity of the interactions between olfactory receptors and volatile molecules and the scarcity of comprehensive olfactory datasets, present unique challenges in this sensory domain. Previous works have explored the relationship between molecular structure and odor descriptors using fully supervised training approaches. However, these methods are data-intensive and poorly generalize due to labeled data scarcity, particularly for rare-class samples. Our study partially tackles the challenges of data scarcity and label skewness through multimodal transfer learning. We investigate the potential of large molecular foundation models trained on extensive unlabeled molecular data to effectively model olfactory perception. Additionally, we explore the integration of different molecular representations, including molecular graphs and text-based SMILES encodings, to achieve data efficiency and generalization of the learned model, particularly on sparsely represented classes. By leveraging complementary representations, we aim to learn robust perceptual features of odorants. However, we observe that traditional methods of combining modalities do not yield substantial gains in high-dimensional skewed label spaces. To address this challenge, we introduce a novel label-balancer technique specifically designed for high-dimensional multi-label and multi-modal training. The label-balancer technique distributes learning objectives across modalities to optimize collaboratively for distinct subsets of labels. Our results suggest that multi-modal transfer features learned using the label-balancer technique are more effective and robust, surpassing the capabilities of traditional uni- or multi-modal approaches, particularly on rare-class samples.

Active metric learning is the problem of incrementally selecting high-utility batches of training data (typically, ordered triplets) to annotate, in order to progressively improve a learned model of a metric over some input domain as rapidly as possible. Standard approaches, which independently assess the informativeness of each triplet in a batch, are susceptible to highly correlated batches with many redundant triplets and hence low overall utility. While a recent work proposes batch-decorrelation strategies for metric learning, they rely on ad hoc heuristics to estimate the correlation between two triplets at a time. We present a novel batch active metric learning method that leverages the Maximum Entropy Principle to learn the least biased estimate of triplet distribution for a given set of prior constraints. To avoid redundancy between triplets, our method collectively selects batches with maximum joint entropy, which simultaneously captures both informativeness and diversity. We take advantage of the submodularity of the joint entropy function to construct a tractable solution using an efficient greedy algorithm based on Gram-Schmidt orthogonalization that is provably (1−\frac1𝑒)-optimal. Our approach is the first batch active metric learning method to define a unified score that balances informativeness and diversity for an entire batch of triplets. Experiments with several real-world datasets demonstrate that our algorithm is robust, generalizes well to different applications and input modalities, and consistently outperforms the state-of-the-art.

We present an active learning strategy for training parametric models of distance metrics, given triplet-based similarity assessments: object xi is more similar to object xj than to xk. In contrast to prior work on class-based learning, where the fundamental goal is classification and any implicit or explicit metric is binary, we focus on perceptual metrics that express the degree of (dis) similarity between objects. We find that standard active learning approaches degrade when annotations are requested for batches of triplets at a time: our studies suggest that correlation among triplets is responsible. In this work, we propose a novel method to decorrelate batches of triplets, that jointly balances informativeness and diversity while decoupling the choice of heuristic for each criterion. Experiments indicate our method is general, adaptable, and outperforms the state-of-the-art.

In order to design haptic icons or build a haptic vocabulary, we require a set of easily distinguishable haptic signals to avoid perceptual ambiguity, which in turn requires a way to accurately estimate the perceptual (dis)similarity of such signals. In this work, we present a novel method to learn such a perceptual metric based on data from human studies. Our method is based on a deep neural network that projects signals to an embedding space where the natural Euclidean distance accurately models the degree of dissimilarity between two signals. The network is trained only on non-numerical comparisons of triplets of signals, using a novel triplet loss that considers both types of triplets that are easy to order (inequality constraints), as well as those that are unorderable/ambiguous (equality constraints). Unlike prior MDS-based non-parametric approaches, our method can be trained on a partial set of comparisons and can embed new haptic signals without retraining the model from scratch. Extensive experimental evaluations show that our method is significantly more effective at modeling perceptual dissimilarity than alternatives.