Raunak Shah

Modern applications commonly leverage large, multi-modal foundation models. These applications often feature complex workflows that demand the storage and usage of similar models in multiple precisions. A straightforward approach is to maintain a separate file for each model precision (e.g., INT8, BF16), which is indeed the approach taken by many model providers such as HuggingFace and Ollama. However, this approach incurs excessive storage costs since a higher precision model (e.g., BF16) is a strict superset of a lower precision model (e.g., INT8) in terms of information. Unfortunately, simply maintaining only the higher-precision model and requiring every user to dynamically convert the model precision is not desirable because every user of lower precision models must pay the cost for model download and precision conversion. In this paper, we present QStore, a unified, lossless compression format for simultaneously storing a model in two (high and low) precisions efficiently. Instead of storing low-precision and high-precision models separately, QStore stores low-precision model and only the residual information needed to reconstruct high-precision models. The size of residual information is significantly smaller than the original high-precision models, thus achieving high savings in storage cost. Moreover, QStore does not compromise the speed of model loading. The low-precision models can be loaded quickly just like before. The high-precision models can also be reconstructed efficiently in memory by merging low-precision data and the residual with QStore’s lightweight decoding logic. We evaluate QStore for compressing multiple precisions of popular foundation models, and show that QStore reduces overall storage footprint by up to 2.2x (45% of the original size) while enabling up to 1.7x and 1.8x faster model saving and loading versus existing approaches.

Enterprise data lakes often suffer from substantial amounts of duplicate and redundant data, with data volumes reaching the scale of terabytes to petabytes. This leads to both increased storage costs and unnecessarily high compute costs for the maintenance of these datasets. In this work we focus on identifying and reducing redundancy in enterprise data lakes by addressing the problem of "dataset containment". To the best of our knowledge, this is one of the first works that addresses table level containment at a large scale. To achieve this, we propose building a dataset containment graph in a scalable manner. Our three-step hierarchical pipeline, R2D2, efficiently identifies all instances of containment by progressively reducing the search space in the data lake via: (i) schema graph builder, (ii) statistical min-max pruning, and (iii) content level pruning. Next, we propose to reduce the storage and maintenance costs due to this redundancy in an optimal manner by deleting datasets, while keeping the customer experience unaffected. We provide an optimization formulation that takes a containment graph as input and identifies redundant datasets that can be deleted (and reconstructed on demand, if necessary), while minimizing the overall storage and access costs and respecting any latency constraints. In contrast to existing modified baselines that are inaccurate or take several days to run, our pipeline can process an enterprise customer data lake at the TB scale in approximately 5 hours, while computing a dataset containment graph that captures all existing containment relationships correctly with very few additional edges compared to the ground truth. We implement our system on Azure Databricks clusters using Apache Spark for enterprise data stored in ADLS Gen2, and on AWS clusters for open-source data. We present theoretical results as well as extensive empirical validation on both enterprise (scale of TB) and open-source datasets (scale of MBs - GBs), which showcases the effectiveness of our pipeline. We discuss important aspects for practical implementation such as dynamic updates of data, scalability and latency concerns, as well as data dependent access patterns. Additionally, we show how each component of our system substantially improves over existing baselines in terms of speed and accuracy of containment detection, as well as how we ensure cost reduction via data deletion.

We study the problem of optimizing data storage and access costs on the cloud while ensuring that the desired performance or latency is unaffected. We first propose an optimizer that optimizes the data placement tier (on the cloud) and the choice of compression schemes to apply, for given data partitions with temporal access predictions. Secondly, we propose a model to learn the compression performance of multiple algorithms across data partitions in different formats to generate compression performance predictions on the fly, as inputs to the optimizer. Thirdly, we propose to approach the data partitioning problem fundamentally differently than the current default in most data lakes where partitioning is in the form of ingestion batches. We propose access pattern aware data partitioning and formulate an optimization problem that optimizes the size and reading costs of partitions subject to access patterns. We study the various optimization problems theoretically as well as empirically, and provide theoretical bounds as well as hardness results. We propose a unified pipeline of cost minimization, called SCOPe that combines the different modules. We extensively compare the performance of our methods with related baselines from the literature on TPC-H data as well as enterprise datasets (ranging from GB to PB in volume) and show that SCOPe substantially improves over the baselines. We show significant cost savings compared to platform baselines, of the order of 50% to 83% on enterprise Data Lake datasets, ranging from terabytes to petabytes in volume.

Solving depth estimation with monocular cameras enables the possibility of widespread use of cameras as low-cost depth estimation sensors in applications such as autonomous driving and robotics. However, learning such a scalable depth estimation model would require a lot of labeled data which is expensive to collect. There are two popular existing approaches which do not require annotated depth maps: (i) using labeled synthetic and unlabeled real data in an adversarial framework to predict more accurate depth, and (ii) unsupervised models which exploit geometric structure across space and time in monocular video frames. Ideally, we would like to leverage features provided by both approaches as they complement each other; however, existing methods do not adequately exploit these additive benefits. We present S3Net, a self-supervised framework which combines these complementary features: we use synthetic and real-world images for training while exploiting geometric, temporal, as well as semantic constraints. Our novel consolidated architecture provides a new state-of-the-art in self-supervised depth estimation using monocular videos. We present a unique way to train this self-supervised framework, and achieve (i) more than 15% improvement over previous synthetic supervised approaches that use domain adaptation and (ii) more than 10% improvement over previous self-supervised approaches which exploit geometric constraints from the real data.

This research paper proposes a COVID-19 monitoring and response system to identify the surge in the volume of patients at hospitals and shortage of critical equipment like ventilators in South-east Asian countries, to understand the burden on health facilities. This can help authorities in these regions with resource planning measures to redirect resources to the regions identified by the model. Due to the lack of publicly available data on the influx of patients in hospitals, or the shortage of equipment, ICU units or hospital beds that regions in these countries might be facing, we leverage Twitter data for gleaning this information. The approach has yielded accurate results for states in India, and we are working on validating the model for the remaining countries so that it can serve as a reliable tool for authorities to monitor the burden on hospitals.

The IIT Kanpur Consulting Group is one of the pioneering research groups in India which focuses on the applications of Machine Learning and Strategy Consulting for social good. The group has been working since 2018 to help social organizations, nonprofits, and government entities in India leverage better insights from their data, with a special emphasis on the healthcare, environmental, and agriculture sectors. The group has worked on critical social problems which India is facing including Polio recurrence, COVID-19, air pollution and agricultural crop damage. This position paper summarises the focus areas and relevant projects which the group has worked on since its establishment, and also highlights the group’s plans for using machine learning to address social problems during the COVID-19 crisis.

The identification and control of human factors in climate change is a rapidly growing concern and robust, real-time air-quality monitoring and forecasting plays a critical role in allowing effective policy formulation and implementation. This paper presents DELFI, a novel deep learning-based mixture model to make effective long-term predictions of Particulate Matter (PM) 2.5 concentrations. A key novelty in DELFI is its multi-scale approach to the forecasting problem. The observation that point predictions are more suitable in the short-term and probabilistic predictions in the long-term allows accurate predictions to be made as much as 24 hours in advance. DELFI incorporates meteorological data as well as pollutant-based features to ensure a robust model that is divided into two parts: (i) a stack of three Long Short-Term Memory (LSTM) networks that perform differential modelling of the same window of past data, and (ii) a fully-connected layer enabling attention to each of the components. Experimental evaluation based on deployment of 13 stations in the Delhi National Capital Region (Delhi-NCR) in India establishes that DELFI offers far superior predictions especially in the long-term as compared to even non-parametric baselines. The Delhi-NCR recorded the 3rd highest PM levels amongst 39 mega-cities across the world during 2011-2015 and DELFI’s performance establishes it as a potential tool for effective long-term forecasting of PM levels to enable public health management and environment protection.