Gaurav Singh

Bipasha Sen, Gaurav Singh, Aditya Agarwal et al. · mila, mit

Neural Radiance Fields (NeRF) have become an increasingly popular representation to capture high-quality appearance and shape of scenes and objects. However, learning generalizable NeRF priors over categories of scenes or objects has been challenging due to the high dimensionality of network weight space. To address the limitations of existing work on generalization, multi-view consistency and to improve quality, we propose HyP-NeRF, a latent conditioning method for learning generalizable category-level NeRF priors using hypernetworks. Rather than using hypernetworks to estimate only the weights of a NeRF, we estimate both the weights and the multi-resolution hash encodings resulting in significant quality gains. To improve quality even further, we incorporate a denoise and finetune strategy that denoises images rendered from NeRFs estimated by the hypernetwork and finetunes it while retaining multiview consistency. These improvements enable us to use HyP-NeRF as a generalizable prior for multiple downstream tasks including NeRF reconstruction from single-view or cluttered scenes and text-to-NeRF. We provide qualitative comparisons and evaluate HyP-NeRF on three tasks: generalization, compression, and retrieval, demonstrating our state-of-the-art results.

Bipasha Sen, Aditya Agarwal, Gaurav Singh et al. · mila, mit

Recovering full 3D shapes from partial observations is a challenging task that has been extensively addressed in the computer vision community. Many deep learning methods tackle this problem by training 3D shape generation networks to learn a prior over the full 3D shapes. In this training regime, the methods expect the inputs to be in a fixed canonical form, without which they fail to learn a valid prior over the 3D shapes. We propose SCARP, a model that performs Shape Completion in ARbitrary Poses. Given a partial pointcloud of an object, SCARP learns a disentangled feature representation of pose and shape by relying on rotationally equivariant pose features and geometric shape features trained using a multi-tasking objective. Unlike existing methods that depend on an external canonicalization, SCARP performs canonicalization, pose estimation, and shape completion in a single network, improving the performance by 45% over the existing baselines. In this work, we use SCARP for improving grasp proposals on tabletop objects. By completing partial tabletop objects directly in their observed poses, SCARP enables a SOTA grasp proposal network improve their proposals by 71.2% on partial shapes. Project page: https://bipashasen.github.io/scarp

Omar Numan, Gaurav Singh, Kazybek Adam et al.

Developing accurate and reliable Compute-In-Memory (CIM) architectures is becoming a key research focus to accelerate Artificial Intelligence (AI) tasks on hardware, particularly Deep Neural Networks (DNNs). In that regard, there has been significant interest in analog and mixed-signal CIM architectures aimed at increasing the efficiency of data storage and computation to handle the massive amount of data needed by DNNs. Specifically, resistive mixed-signal CIM cores are pushed by recent progresses in emerging Non-Volatile Memory (eNVM) solutions. Yet, mixed-signal CIM computing cores still face several integration and reliability challenges that hinder their large-scale adoption into end-to-end AI computing systems. In terms of integration, resistive and eNVM-based CIM cores need to be integrated with a control processor to realize end-to-end AI acceleration. Moreover, SRAM-based CIM architectures are still more efficient and easier to program than their eNVM counterparts. In terms of reliability, analog circuits are more susceptible to variations, leading to computation errors and degraded accuracy. This work addresses these two challenges by proposing a self-calibrated mixed-signal CIM accelerator SoC, fabricated in 22-nm FDSOI technology. The integration is facilitated by (1) the CIM architecture, combining the density and ease of SRAM-based weight storage with multi-bit computation using linear resistors, and (2) an open-source programming and testing strategy for CIM systems. The accuracy and reliability are enabled through an automated RISC-V controlled on-chip calibration, allowing us to improve the compute SNR by 25 to 45% across multiple columns to reach 18-24 dB. To showcase further integration possibilities, we show how our proof-of-concept SoC can be extended to recent high-density linear resistor technologies for enhanced computing performance.

Talia Gershon, Seetharami Seelam, Brian Belgodere et al.

AI Infrastructure plays a key role in the speed and cost-competitiveness of developing and deploying advanced AI models. The current demand for powerful AI infrastructure for model training is driven by the emergence of generative AI and foundational models, where on occasion thousands of GPUs must cooperate on a single training job for the model to be trained in a reasonable time. Delivering efficient and high-performing AI training requires an end-to-end solution that combines hardware, software and holistic telemetry to cater for multiple types of AI workloads. In this report, we describe IBM's hybrid cloud infrastructure that powers our generative AI model development. This infrastructure includes (1) Vela: an AI-optimized supercomputing capability directly integrated into the IBM Cloud, delivering scalable, dynamic, multi-tenant and geographically distributed infrastructure for large-scale model training and other AI workflow steps and (2) Blue Vela: a large-scale, purpose-built, on-premises hosting environment that is optimized to support our largest and most ambitious AI model training tasks. Vela provides IBM with the dual benefit of high performance for internal use along with the flexibility to adapt to an evolving commercial landscape. Blue Vela provides us with the benefits of rapid development of our largest and most ambitious models, as well as future-proofing against the evolving model landscape in the industry. Taken together, they provide IBM with the ability to rapidly innovate in the development of both AI models and commercial offerings.

Pavan Karke, Kushal Shah, Gaurav Singh et al.

The problem of object pose and shape estimation has seen key advancements lately. Encoder-decoder (e.g., SAM3D, LRM, CRISP) and diffusion-based models (e.g., InstantMesh, Zero123, SceneComplete) have shown category-agnostic shape encoding capacity and open-set generalizability. In this work, we ask the question: Are the object pose and shape estimation methods mature enough, such that when used with antipodal grasp sampling, can outperform the end-to-end grasp synthesis methods? We explore this question in detail by scoping our study to parallel jaw grippers, 7-DoF grasps, and single-view RGB(-D) image as input. We implement and compare a state-of-the-art, end-to-end grasp synthesis method and three modular methods, which first estimate the object pose and shape for all objects in the scene, and generate grasps using antipodal sampling. We observe that the modular methods outperform the end-to-end method in all our experiments. The modular methods are able to synthesize plenty of grasps, even for small objects, where the end-to-end methods fail. The effectiveness of the modular methods is contingent on the accuracy of the pose and shape estimation, and suffers partial degradation in cluttered scenes - a limitation of the existing pose and shape estimation methods. We also analyze the failure modes and run-times for the three modular methods, which use two different ways of object pose and shape estimation: one based on an encoder-decoder model, while another a diffusion model. Finally, we demonstrate that the single-view object pose and shape estimation methods can be augmented with vision-language models to yield language-conditioned grasps from just single-view RGB-D image as input. We notice comparable performance to the state-of-the-art LERF-TOGO baseline.

Md Maidul Islam, Md Omar Faruque, Joshua Butterfield et al.

Power quality (PQ) events are recorded by PQ meters whenever anomalous events are detected on the power grid. Using neural networks with machine learning can aid in accurately classifying the recorded waveforms and help power system engineers diagnose and rectify the root causes of problems. However, many of the waveforms captured during a disturbance in the power system need to be labeled for supervised learning, leaving a large number of data recordings for engineers to process manually or go unseen. This paper presents an autoencoder and K-means clustering-based unsupervised technique that can be used to cluster PQ events into categories like sag, interruption, transients, normal, and harmonic distortion to enable filtering of anomalous waveforms from recurring or normal waveforms. The method is demonstrated using three-phase, field-obtained voltage waveforms recorded in a distribution grid. First, a convolutional autoencoder compresses the input signals into a set of lower feature dimensions which, after further processing, is passed to the K-means algorithm to identify data clusters. Using a small, labeled dataset, numerical labels are then assigned to events based on a cosine similarity analysis. Finally, the study analyzes the clusters using the t-distributed stochastic neighbor embedding (t-SNE) visualization tool, demonstrating that the technique can help investigate a large number of captured events in a quick manner.

Paridhi Singh, Gaurav Singh, Arun Kumar

Determining accurate bird's eye view (BEV) positions of objects and tracks in a scene is vital for various perception tasks including object interactions mapping, scenario extraction etc., however, the level of supervision required to accomplish that is extremely challenging to procure. We propose a light-weight, weakly supervised method to estimate 3D position of objects by jointly learning to regress the 2D object detections and scene's depth prediction in a single feed-forward pass of a network. Our proposed method extends a center-point based single-shot object detector, and introduces a novel object representation where each object is modeled as a BEV point spatio-temporally, without the need of any 3D or BEV annotations for training and LiDAR data at query time. The approach leverages readily available 2D object supervision along with LiDAR point clouds (used only during training) to jointly train a single network, that learns to predict 2D object detection alongside the whole scene's depth, to spatio-temporally model object tracks as points in BEV. The proposed method is computationally over $\sim$10x efficient compared to recent SOTA approaches while achieving comparable accuracies on KITTI tracking benchmark.

Gaurav Singh, Sanket Kalwar, Md Faizal Karim et al. · mit

Efficiently generating grasp poses tailored to specific regions of an object is vital for various robotic manipulation tasks, especially in a dual-arm setup. This scenario presents a significant challenge due to the complex geometries involved, requiring a deep understanding of the local geometry to generate grasps efficiently on the specified constrained regions. Existing methods only explore settings involving table-top/small objects and require augmented datasets to train, limiting their performance on complex objects. We propose CGDF: Constrained Grasp Diffusion Fields, a diffusion-based grasp generative model that generalizes to objects with arbitrary geometries, as well as generates dense grasps on the target regions. CGDF uses a part-guided diffusion approach that enables it to get high sample efficiency in constrained grasping without explicitly training on massive constraint-augmented datasets. We provide qualitative and quantitative comparisons using analytical metrics and in simulation, in both unconstrained and constrained settings to show that our method can generalize to generate stable grasps on complex objects, especially useful for dual-arm manipulation settings, while existing methods struggle to do so.

Gaurav Singh, Kavitesh Kumar Bali

This paper explores the seamless integration of Generative AI (GenAI) and Evolutionary Algorithms (EAs) within the domain of large-scale multi-objective optimization. Focusing on the transformative role of Large Language Models (LLMs), our study investigates the potential of LLM-Assisted Inference to automate and enhance decision-making processes. Specifically, we highlight its effectiveness in illuminating key decision variables in evolutionarily optimized solutions while articulating contextual trade-offs. Tailored to address the challenges inherent in inferring complex multi-objective optimization solutions at scale, our approach emphasizes the adaptive nature of LLMs, allowing them to provide nuanced explanations and align their language with diverse stakeholder expertise levels and domain preferences. Empirical studies underscore the practical applicability and impact of LLM-Assisted Inference in real-world decision-making scenarios.

Lingyun Yao, Martin Trapp, Jelin Leslin et al.

Probabilistic circuits (PCs) offer a promising avenue to perform embedded reasoning under uncertainty. They support efficient and exact computation of various probabilistic inference tasks by design. Hence, hardware-efficient computation of PCs is highly interesting for edge computing applications. As computations in PCs are based on arithmetic with probability values, they are typically performed in the log domain to avoid underflow. Unfortunately, performing the log operation on hardware is costly. Hence, prior work has focused on computations in the linear domain, resulting in high resolution and energy requirements. This work proposes the first dedicated approximate computing framework for PCs that allows for low-resolution logarithm computations. We leverage Addition As Int, resulting in linear PC computation with simple hardware elements. Further, we provide a theoretical approximation error analysis and present an error compensation mechanism. Empirically, our method obtains up to 357x and 649x energy reduction on custom hardware for evidence and MAP queries respectively with little or no computational error.

Radhika Dua, Nikita Saxena, Aditi Agarwal et al.

Agricultural landscapes are quite complex, especially in the Global South where fields are smaller, and agricultural practices are more varied. In this paper we report on our progress in digitizing the agricultural landscape (natural and man-made) in our study region of India. We use high resolution imagery and a UNet style segmentation model to generate the first of its kind national-scale multi-class panoptic segmentation output. Through this work we have been able to identify individual fields across 151.7M hectares, and delineating key features such as water resources and vegetation. We share how this output was validated by our team and externally by downstream users, including some sample use cases that can lead to targeted data driven decision making. We believe this dataset will contribute towards digitizing agriculture by generating the foundational baselayer.

Abhishek Dey, Saurabh Srivastava, Gaurav Singh et al.

This paper presents PICO-TINYML-BENCHMARK, a modular and platform-agnostic framework for benchmarking the real-time performance of TinyML models on resource-constrained embedded systems. Evaluating key metrics such as inference latency, CPU utilization, memory efficiency, and prediction stability, the framework provides insights into computational trade-offs and platform-specific optimizations. We benchmark three representative TinyML models -- Gesture Classification, Keyword Spotting, and MobileNet V2 -- on two widely adopted platforms, BeagleBone AI64 and Raspberry Pi 4, using real-world datasets. Results reveal critical trade-offs: the BeagleBone AI64 demonstrates consistent inference latency for AI-specific tasks, while the Raspberry Pi 4 excels in resource efficiency and cost-effectiveness. These findings offer actionable guidance for optimizing TinyML deployments, bridging the gap between theoretical advancements and practical applications in embedded systems.

Gaurav Singh

Pneumonia, particularly when induced by diseases like COVID-19, remains a critical global health challenge requiring rapid and accurate diagnosis. This study presents a comprehensive comparison of traditional machine learning and state-of-the-art deep learning approaches for automated pneumonia detection using chest X-rays (CXRs). We evaluate multiple methodologies, ranging from conventional machine learning techniques (PCA-based clustering, Logistic Regression, and Support Vector Classification) to advanced deep learning architectures including Convolutional Neural Networks (Modified LeNet, DenseNet-121) and various Vision Transformer (ViT) implementations (Deep-ViT, Compact Convolutional Transformer, and Cross-ViT). Using a dataset of 5,856 pediatric CXR images, we demonstrate that Vision Transformers, particularly the Cross-ViT architecture, achieve superior performance with 88.25% accuracy and 99.42% recall, surpassing traditional CNN approaches. Our analysis reveals that architectural choices impact performance more significantly than model size, with Cross-ViT's 75M parameters outperforming larger models. The study also addresses practical considerations including computational efficiency, training requirements, and the critical balance between precision and recall in medical diagnostics. Our findings suggest that Vision Transformers offer a promising direction for automated pneumonia detection, potentially enabling more rapid and accurate diagnosis during health crises.

Ishan Deshpande, Amandeep Kaur Reehal, Chandan Nath et al.

Accurate, timely, and farm-level crop type information is paramount for national food security, agricultural policy formulation, and economic planning, particularly in agriculturally significant nations like India. While remote sensing and machine learning have become vital tools for crop monitoring, existing approaches often grapple with challenges such as limited geographical scalability, restricted crop type coverage, the complexities of mixed-pixel and heterogeneous landscapes, and crucially, the robust in-season identification essential for proactive decision-making. We present a framework designed to address the critical data gaps for targeted data driven decision making which generates farm-level, in-season, multi-crop identification at national scale (India) using deep learning. Our methodology leverages the strengths of Sentinel-1 and Sentinel-2 satellite imagery, integrated with national-scale farm boundary data. The model successfully identifies 12 major crops (which collectively account for nearly 90% of India's total cultivated area showing an agreement with national crop census 2023-24 of 94% in winter, and 75% in monsoon season). Our approach incorporates an automated season detection algorithm, which estimates crop sowing and harvest periods. This allows for reliable crop identification as early as two months into the growing season and facilitates rigorous in-season performance evaluation. Furthermore, we have engineered a highly scalable inference pipeline, culminating in what is, to our knowledge, the first pan-India, in-season, farm-level crop type data product. The system's effectiveness and scalability are demonstrated through robust validation against national agricultural statistics, showcasing its potential to deliver actionable, data-driven insights for transformative agricultural monitoring and management across India.

Saurabh Srivastava, Gaurav Singh, Shou Matsumoto et al.

In this work, we present the first dataset, MailEx, for performing event extraction from conversational email threads. To this end, we first proposed a new taxonomy covering 10 event types and 76 arguments in the email domain. Our final dataset includes 1.5K email threads and ~4K emails, which are annotated with totally ~8K event instances. To understand the task challenges, we conducted a series of experiments comparing three types of approaches, i.e., fine-tuned sequence labeling, fine-tuned generative extraction, and few-shot in-context learning. Our results showed that the task of email event extraction is far from being addressed, due to challenges lying in, e.g., extracting non-continuous, shared trigger spans, extracting non-named entity arguments, and modeling the email conversational history. Our work thus suggests more future investigations in this domain-specific event extraction task.

Gaurav Singh, Siffi Singh, Joshua Wong et al.

Relation Extraction (RE) from tables is the task of identifying relations between pairs of columns of a table. Generally, RE models for this task require labelled tables for training. These labelled tables can also be generated artificially from a Knowledge Graph (KG), which makes the cost to acquire them much lower in comparison to manual annotations. However, unlike real tables, these synthetic tables lack associated metadata, such as, column-headers, captions, etc; this is because synthetic tables are created out of KGs that do not store such metadata. Meanwhile, previous works have shown that metadata is important for accurate RE from tables. To address this issue, we propose methods to artificially create some of this metadata for synthetic tables. Afterward, we experiment with a BERT-based model, in line with recently published works, that takes as input a combination of proposed artificial metadata and table content. Our empirical results show that this leads to an improvement of 9\%-45\% in F1 score, in absolute terms, over 2 tabular datasets.

Gaurav Singh

This paper discusses the results obtained for different techniques applied for performing the sentiment analysis of social media (Twitter) code-mixed text written in Hinglish. The various stages involved in performing the sentiment analysis were data consolidation, data cleaning, data transformation and modelling. Various data cleaning techniques were applied, data was cleaned in five iterations and the results of experiments conducted were noted after each iteration. Data was transformed using count vectorizer, one hot vectorizer, tf-idf vectorizer, doc2vec, word2vec and fasttext embeddings. The models were created using various machine learning algorithms such as SVM, KNN, Decision Trees, Random Forests, Naive Bayes, Logistic Regression, and ensemble voting classifiers. The data was obtained from a task on Codalab competition website which was listed as Task:9 on the Semeval-2020 competition website. The models created were evaluated using the F1-score (macro). The best F1-score of 69.07 was achieved using ensemble voting classifier.

Gaurav Singh

This paper discusses the design of the system used for providing a solution for the problem given at SemEval-2020 Task 9 where sentiment analysis of code-mixed language Hindi and English needed to be performed. This system uses Weka as a tool for providing the classifier for the classification of tweets and python is used for loading the data from the files provided and cleaning it. Only part of the training data was provided to the system for classifying the tweets in the test data set on which evaluation of the system was done. The system performance was assessed using the official competition evaluation metric F1-score. Classifier was trained on two sets of training data which resulted in F1 scores of 0.4972 and 0.5316.

Gaurav Singh, Zahra Sabet, John Shawe-Taylor et al.

Biomedical text tagging systems are plagued by the dearth of labeled training data. There have been recent attempts at using pre-trained encoders to deal with this issue. Pre-trained encoder provides representation of the input text which is then fed to task-specific layers for classification. The entire network is fine-tuned on the labeled data from the target task. Unfortunately, a low-resource biomedical task often has too few labeled instances for satisfactory fine-tuning. Also, if the label space is large, it contains few or no labeled instances for majority of the labels. Most biomedical tagging systems treat labels as indexes, ignoring the fact that these labels are often concepts expressed in natural language e.g. `Appearance of lesion on brain imaging'. To address these issues, we propose constructing extra labeled instances using label-text (i.e. label's name) as input for the corresponding label-index (i.e. label's index). In fact, we propose a number of strategies for manufacturing multiple artificial labeled instances from a single label. The network is then fine-tuned on a combination of real and these newly constructed artificial labeled instances. We evaluate the proposed approach on an important low-resource biomedical task called \textit{PICO annotation}, which requires tagging raw text describing clinical trials with labels corresponding to different aspects of the trial i.e. PICO (Population, Intervention/Control, Outcome) characteristics of the trial. Our empirical results show that the proposed method achieves a new state-of-the-art performance for PICO annotation with very significant improvements over competitive baselines.

Gaurav Singh, Parminder Bhatia

Relation Extraction (RE) aims to label relations between groups of marked entities in raw text. Most current RE models learn context-aware representations of the target entities that are then used to establish relation between them. This works well for intra-sentence RE and we call them first-order relations. However, this methodology can sometimes fail to capture complex and long dependencies. To address this, we hypothesize that at times two target entities can be explicitly connected via a context token. We refer to such indirect relations as second-order relations and describe an efficient implementation for computing them. These second-order relation scores are then combined with first-order relation scores. Our empirical results show that the proposed method leads to state-of-the-art performance over two biomedical datasets.

Gaurav Singh, James Thomas, Iain J. Marshall et al.

We propose a model for tagging unstructured texts with an arbitrary number of terms drawn from a tree-structured vocabulary (i.e., an ontology). We treat this as a special case of sequence-to-sequence learning in which the decoder begins at the root node of an ontological tree and recursively elects to expand child nodes as a function of the input text, the current node, and the latent decoder state. In our experiments the proposed method outperforms state-of-the-art approaches on the important task of automatically assigning MeSH terms to biomedical abstracts.

Gaurav Singh, John Shawe-Taylor

Deep neural networks have gained tremendous popularity in last few years. They have been applied for the task of classification in almost every domain. Despite the success, deep networks can be incredibly slow to train for even moderate sized models on sufficiently large datasets. Additionally, these networks require large amounts of data to be able to generalize. The importance of speeding up convergence, and generalization in deep networks can not be overstated. In this work, we develop an optimization algorithm based on generalized-optimal updates derived from minibatches that lead to faster convergence. Towards the end, we demonstrate on two benchmark datasets that the proposed method achieves two orders of magnitude speed up over traditional back-propagation, and is more robust to noise/over-fitting.

Gaurav Singh, James Thomas, John Shawe-Taylor

Systematic reviews are essential to summarizing the results of different clinical and social science studies. The first step in a systematic review task is to identify all the studies relevant to the review. The task of identifying relevant studies for a given systematic review is usually performed manually, and as a result, involves substantial amounts of expensive human resource. Lately, there have been some attempts to reduce this manual effort using active learning. In this work, we build upon some such existing techniques, and validate by experimenting on a larger and comprehensive dataset than has been attempted until now. Our experiments provide insights on the use of different feature extraction models for different disciplines. More importantly, we identify that a naive active learning based screening process is biased in favour of selecting similar documents. We aimed to improve the performance of the screening process using a novel active learning algorithm with success. Additionally, we propose a mechanism to choose the best feature extraction method for a given review.

Sandra Mitrović, Gaurav Singh

There is an abundance of temporal and non-temporal data in banking (and other industries), but such temporal activity data can not be used directly with classical machine learning models. In this work, we perform extensive feature extraction from the temporal user activity data in an attempt to predict user visits to different branches and credit card up-selling utilizing user information and the corresponding activity data, as part of \emph{ECML/PKDD Discovery Challenge 2016 on Bank Card Usage Analysis}. Our solution ranked \nth{4} for \emph{Task 1} and achieved an AUC of \textbf{$0.7056$} for \emph{Task 2} on public leaderboard.

Gaurav Singh, Fabrizio Silvestri, John Shawe-Taylor

In a traditional setting, classifiers are trained to approximate a target function $f:X \rightarrow Y$ where at least a sample for each $y \in Y$ is presented to the training algorithm. In a zero-shot setting we have a subset of the labels $\hat{Y} \subset Y$ for which we do not observe any corresponding training instance. Still, the function $f$ that we train must be able to correctly assign labels also on $\hat{Y}$. In practice, zero-shot problems are very important especially when the label set is large and the cost of editorially label samples for all possible values in the label set might be prohibitively high. Most recent approaches to zero-shot learning are based on finding and exploiting relationships between labels using semantic embeddings. We show in this paper that semantic embeddings, despite being very good at capturing relationships between labels, are not very good at capturing the relationships among labels in a data-dependent manner. For this reason, we propose a novel two-step process for learning a zero-shot classifier. In the first step, we learn what we call a \emph{property embedding space} capturing the "\emph{learnable}" features of the label set. Then, we exploit the learned properties in order to reduce the generalization error for a linear nearest neighbor-based classifier.

Gaurav Singh, Benjamin Piwowarski

We present a novel method for efficiently searching top-k neighbors for documents represented in high dimensional space of terms based on the cosine similarity. Mostly, documents are stored as bag-of-words tf-idf representation. One of the most used ways of computing similarity between a pair of documents is cosine similarity between the vector representations, but cosine similarity is not a metric distance measure as it doesn't follow triangle inequality, therefore most metric searching methods can not be applied directly. We propose an efficient method for indexing documents using a pivot tree that leads to efficient retrieval. We also study the relation between precision and efficiency for the proposed method and compare it with a state of the art in the area of document searching based on inner product.