Abstract

Graph Neural Networks (GNNs) have emerged as powerful tools for learning representations of graph-structured data, demonstrating remarkable performance across various tasks. Recognising their importance, there has been extensive research focused on explaining GNN predictions, aiming to enhance their interpretability and trustworthiness. However, GNNs and their explainers face a notable challenge: graphs are primarily designed to model pair-wise relationships between nodes, which can make it tough to capture higher-order, multi-node interactions. This characteristic can pose difficulties for existing explainers in fully representing multi-node relationships. To address this gap, we present Framework For Higher-Order Representations In Graph Explanations (FORGE), a framework that enables graph explainers to capture such interactions by incorporating higher-order structures, resulting in more accurate and faithful explanations. Extensive evaluation shows that on average real-world datasets from the GraphXAI benchmark and synthetic datasets across various graph explainers, FORGE improves average explanation accuracy by 1.9x and 2.25x, respectively. We perform ablation studies to confirm the importance of higher-order relations in improving explanations, while our scalability analysis demonstrates FORGE's efficacy on large graphs.

Abstract

One of the key tasks in graph learning is node classification. While Graph neural networks have been used for various applications, their adaptivity to reject option setting is not previously explored. In this paper, we propose NCwR, a novel approach to node classification in Graph Neural Networks (GNNs) with an integrated reject option, which allows the model to abstain from making predictions when uncertainty is high. We propose both cost-based and coverage-based methods for classification with abstention in node classification setting using GNNs. We perform experiments using our method on three standard citation network datasets Cora, Citeseer and Pubmed and compare with relevant baselines. We also model the Legal judgment prediction problem on ILDC dataset as a node classification problem where nodes represent legal cases and edges represent citations. We further interpret the model by analyzing the cases that the model abstains from predicting by visualizing which part of the input features influenced this decision.

Abstract

The significant improvements in point cloud representation learning have increased its applicability in many real-life applications, resulting in the need for lightweight, better-performing models. One widely proposed efficient method is knowledge distillation, where a lightweight model uses knowledge from large models. Very few works exist on distilling the knowledge for point clouds. Most of the work focuses on cross-modal-based approaches that make the method expensive to train. This paper proposes PointKAD, an adversarial knowledge distillation framework for point cloud-based tasks. PointKAD includes adversarial feature distillation and response distillation with the help of discriminators to extract and distill the representation of feature maps and logits. We conduct extensive experimental studies on both synthetic (ModelNet40) and real (ScanObjectNN) datasets to show that PointKAD achieves state-of-the-art results compared to the existing knowledge distillation methods for point cloud classification. Additionally, we present results on the part segmentation task, highlighting the efficacy of the PointKAD framework. Our experiments further reveal that PointKAD is capable of transferring knowledge across different tasks and datasets, showcasing its versatility. Furthermore, we demonstrate that PointKAD can be applied to a cross-modal training setup, achieving competitive performance with cross-modal-based point cloud methods for classification.

Abstract

Text driven diffusion models have shown remarkable capabilities in editing images. However, when editing 3D scenes, existing works mostly rely on training a NeRF for 3D editing. Recent NeRF editing methods leverages edit operations by deploying 2D diffusion models and project these edits into 3D space. They require strong positional priors alongside text prompt to identify the edit location. These methods are operational on small 3D scenes and are more generalized to particular scene. They require training for each specific edit and cannot be exploited in real-time edits. To address these limitations, we propose a novel method, FreeEdit, to make edits in training free manner using mesh representations as a substitute for NeRF. Training-free methods are now a possibility because of the advances in foundation model’s space. We leverage these models to bring a training-free alternative and introduce solutions for insertion, replacement and deletion. We consider insertion, replacement and deletion as basic blocks for performing intricate edits with certain combinations of these operations. Given a text prompt and a 3D scene, our model is capable of identifying what object should be inserted/replaced or deleted and location where edit should be performed. We also introduce a novel algorithm as part of FreeEdit to find the optimal location on grounding object for placement. We evaluate our model by comparing it with baseline models on a wide range of scenes using quantitative and qualitative metrics and showcase the merits of our method with respect to others.

Abstract

Autonomous Underwater Vehicles (AUVs) play a vital role in explor- ing and mapping underwater environments. However, the presence of nadir gaps, or blind zones, in commercial AUVs can lead to unex- plored areas during mission execution, limiting their effectiveness. Our work addresses the challenges of path planning in the presence of nadir gaps and presents scalable coverage strategies for AUVs minimizing either the mission completion time or the total number of turns performed while ensuring complete exploration, eliminating the risk of leaving critical areas unexplored. We provide provably complete strategies and perform extensive simulations on diverse input configurations based on real-world instances to demonstrate the efficacy of our strategies.

Abstract

Given the current emphasis on maintaining and inspecting high-rise buildings, conventional inspection approaches are costly, slow, error-prone, and labor-intensive due to manual processes and lack of automation. In this paper, we provide an automated, periodic, accurate and economical solution for the inspection of such buildings on real-world images. We propose a novel end-to-end integrated autonomous pipeline for building inspection which consists of three modules: i) Autonomous Drone Navigation, ii) Façade Detection, and iii) Model Construction. Our first module computes a collision-free trajectory for the UAV around the building for surveillance. The images captured in this step are used for façade detection and 3D building model construction. The façade detection module is a deep learning-based object detection method which detects cracks. Finally, the model construction module focuses on reconstructing a 3D model of a building from captured images to mark the corresponding cracks on the 3D model for efficient and accurate inferences from the inspection. We conduct experiments for each module, including collision avoidance for drone navigation, façade detection, model construction and mapping. Our experimental analysis shows the promising performance of i) our crack detection model with a precision and recall of 0.95 and mAP score of 0.96; ii) our 3D reconstruction method includes finer details of the building without having additional information on the sequence of images; and iii) our 2D-3D mapping to compute the original location/world coordinates of cracks for a building.

Abstract

Recently, the fundamental problem of unsupervised domain adaptation (UDA) on 3D point clouds has been motivated by a wide variety of applications in robotics, virtual reality, and scene understanding, to name a few. The point cloud data acquisition procedures manifest themselves as significant domain discrepancies and geometric variations among both similar and dissimilar classes. The standard domain adaptation methods developed for images do not directly translate to point cloud data because of their complex geometric nature. To address this challenge, we leverage the idea of multimodality and alignment between distributions. We propose a new UDA architecture for point cloud classification that benefits from multimodal contrastive learning to get better class separation in both domains individually. Further, the use of optimal transport (OT) aims at learning source and target data distributions jointly to reduce the cross-domain shift and provide a better alignment. We conduct a comprehensive empirical study on PointDA-10 and GraspNetPC-10 and show that our method achieves state-of-the-art performance on GraspNetPC-10 (with ≈ 4-12% margin) and best average performance on PointDA-10. Our ablation studies and decision boundary analysis also validate the significance of our contrastive learning module and OT alig

Abstract

Learning embeddings of any data largely depends on the ability of the target space to capture semantic relations. The widely used Euclidean space, where embeddings are represented as point vectors, is known to be lacking in its potential to exploit complex structures and relations. Contrary to standard Euclidean embeddings, in this work, we embed point clouds as discrete probability distributions in Wasserstein space. We build a contrastive learning setup to learn Wasserstein embeddings that can be used as a pre-training method with or without supervision towards any downstream task. We show that the features captured by Wasserstein embeddings are better in preserving the point cloud geometry, including both global and local information, thus resulting in improved quality embeddings. We perform exhaustive experiments and demonstrate the effectiveness of our method for point cloud classification, transfer learning, segmentation, and interpolation tasks over multiple datasets including synthetic and realworld objects. We also compare against recent methods that use Wasserstein space and show that our method outperforms them in all downstream tasks. Additionally, our study reveals a promising interpretation of capturing critical points of point clouds that makes our proposed method self-explainable.

Abstract

Commonsense question-answering (QA) methods combine the power of pre-trained Language Models (LM) with the reasoning provided by Knowledge Graphs (KG). A typical approach collects nodes relevant to the QA pair from a KG to form a Working Graph (WG) followed by reasoning using Graph Neural Networks (GNNs). This faces two major challenges: (i) it is difficult to capture all the information from the QA in the WG, and (ii) the WG contains some irrelevant nodes from the KG. To address these, we propose GrapeQA with two simple improvements on the WG: (i) Prominent Entities for Graph Augmentation identifies relevant text chunks from the QA pair and augments the WG with corresponding latent representations from the LM, and (ii) ContextAware Node Pruning removes nodes that are less relevant to the QA pair. We evaluate our results on OpenBookQA, CommonsenseQA and MedQA-USMLE and see that GrapeQA shows consistent improvements over its LM + KG predecessor (QA-GNN in particular) and large improvements on OpenBookQA.

Abstract

Writing a good job description is an important step in the online recruitment process to hire the best candidates. Most recruiters forget to include some relevant skills in the job description. These missing skills affect the performance of recruitment tasks such as job suggestions, job search, candidate recommendations, etc. Existing approaches are limited to contextual modelling, do not exploit inter-relational structures like job-job and job-skill relationships, and are not scalable. In this paper, we exploit these structural relationships using a graph-based approach. We propose a novel skill prediction framework called JobXMLC, which uses graph neural networks with skill attention to predict missing skills using job descriptions. JobXMLC enables joint learning over a job-skill graph consisting of 22.8K entities (jobs and skills) and 650K relationships. We experiment with real-world recruitment datasets to evaluate our proposed approach. We train JobXMLC on 20, 298 jobs and 2, 548 skills within 30 minutes on a single GPU machine. JobXMLC outperforms the state-of-the-art approaches by 6% on precision and 3% on recall. JobXMLC is 18X faster for training tasks and up to 634X faster in skill prediction on benchmark datasets enabling JobXMLC to scale up on larger datasets. We have made our code and dataset public at https://precog.iiit.ac.in/resources.html.