arXiv subjects

Ankur Jain

Publications and source records attributed to Ankur Jain.

9 recordsLinked to original sources

arXiv2026

A finite element-based eigenvalue analysis for predicting thermal runaway in Li-ion battery packs

Thermal runaway remains a critical safety concern in battery systems and other thermally active devices, necessitating reliable methods for predicting the conditions under which thermal runaway may occur. In this work, we develop a finite element framework for assessing thermal stability in systems undergoing transient heat conduction with temperature-dependent internal heat generation. The formulation leads naturally to a generalized eigenvalue problem, wherein the sign of the smallest eigenvalue provides a direct criterion for determining the onset of thermal runaway. This approach enables direct prediction of thermal runaway thresholds in geometrically complicated problems without requiring computationally expensive transient analysis. The proposed methodology is validated through comparison with analytical solutions for a cylindrical Li-ion cell, as well as with a pack of cylindrical cells, demonstrating excellent agreement in both cases. Based on this model, the influence of material properties, boundary conditions, geometric parameters, and spatially varying heat generation on stability limits is examined. A key advantage of this formulation is its effectiveness even for complex geometries where analytical methods become impractical. By providing a systematic and computationally efficient means to identify stability thresholds, the present work offers a practical tool for the thermal design and safety assessment of battery systems.

math-ph↗

arXiv2025

Exact analysis of potential flow past bodies of irregular shapes

Fluid flow past one or more solid bodies is a fundamental problem of much practical importance. Standard solutions of simplified problems involving incompressible inviscid irrotational flow past common geometries such as circular cylinders and airfoils are commonly available. This work presents exact analysis of a potential flow problem involving fluid flow past one or more bodies of irregular shapes. The problem is solved by expressing the shape of each body using Heaviside functions, and writing the potential function as an eigenfunction-based series. Using the properties of Heaviside functions, the series coefficients are determined by deriving a set of linear algebraic equations that govern the coefficients. Benchmarking of the analytical technique against well-known solutions of standard problems is carried out, showing excellent agreement. Good agreement with past work on the specific problem of potential flow past multiple circular cylinders further establishes the accuracy of the analytical technique. Illustrative problems of flow past complicated geometries are solved. Implementation aspects and limitations of the analytical technique are discussed.

physics.flu-dyn↗

arXiv2025

Click A, Buy B: Rethinking Conversion Attribution in E- Commerce Recommendations

User journeys in e-commerce routinely violate the one-to-one assumption that a clicked item on an advertising platform is the same item later purchased on the merchant's website/app. For a significant number of converting sessions on our platform, users click product A but buy product B -- the Click A, Buy B (CABB) phenomenon. Training recommendation models on raw click-conversion pairs therefore rewards items that merely correlate with purchases, leading to biased learning and sub-optimal conversion rates. We reframe conversion prediction as a multi-task problem with separate heads for Click A Buy A (CABA) and Click A Buy B (CABB). To isolate informative CABB conversions from unrelated CABB conversions, we introduce a taxonomy-aware collaborative filtering weighting scheme where each product is first mapped to a leaf node in a product taxonomy, and a category-to-category similarity matrix is learned from large-scale co-engagement logs. This weighting amplifies pairs that reflect genuine substitutable or complementary relations while down-weighting coincidental cross-category purchases. Offline evaluation on e-commerce sessions reduces normalized entropy by 13.9% versus a last-click attribution baseline. An online A/B test on live traffic shows +0.25% gains in the primary business metric.

cs.IR↗

arXiv2025

Apple Intelligence Foundation Language Models: Tech Report 2025

We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon through architectural innovations such as KV-cache sharing and 2-bit quantization-aware training; and ii a scalable server model built on a novel Parallel-Track Mixture-of-Experts PT-MoE transformer that combines track parallelism, mixture-of-experts sparse computation, and interleaved global-local attention to deliver high quality with competitive cost on Apple's Private Cloud Compute platform. Both models are trained on large-scale multilingual and multimodal datasets sourced via responsible web crawling, licensed corpora, and high-quality synthetic data, then further refined with supervised fine-tuning and reinforcement learning on a new asynchronous platform. The resulting models support several additional languages while understanding images and executing tool calls. In public benchmarks and human evaluations, both the server model and the on-device model match or surpass comparably sized open baselines. A new Swift-centric Foundation Models framework exposes guided generation, constrained tool calling, and LoRA adapter fine-tuning, allowing developers to integrate these capabilities with a few lines of code. The latest advancements in Apple Intelligence models are grounded in our Responsible AI approach with safeguards like content filtering and locale-specific evaluation, as well as our commitment to protecting our users' privacy with innovations like Private Cloud Compute.

cs.LG↗

arXiv2024

Apple Intelligence Foundation Language Models

We present foundation language models developed to power Apple Intelligence features, including a ~3 billion parameter model designed to run efficiently on devices and a large server-based language model designed for Private Cloud Compute. These models are designed to perform a wide range of tasks efficiently, accurately, and responsibly. This report describes the model architecture, the data used to train the model, the training process, how the models are optimized for inference, and the evaluation results. We highlight our focus on Responsible AI and how the principles are applied throughout the model development.

cs.AI↗

arXiv2024

MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training

In this work, we discuss building performant Multimodal Large Language Models (MLLMs). In particular, we study the importance of various architecture components and data choices. Through careful and comprehensive ablations of the image encoder, the vision language connector, and various pre-training data choices, we identified several crucial design lessons. For example, we demonstrate that for large-scale multimodal pre-training using a careful mix of image-caption, interleaved image-text, and text-only data is crucial for achieving state-of-the-art (SOTA) few-shot results across multiple benchmarks, compared to other published pre-training results. Further, we show that the image encoder together with image resolution and the image token count has substantial impact, while the vision-language connector design is of comparatively negligible importance. By scaling up the presented recipe, we build MM1, a family of multimodal models up to 30B parameters, including both dense models and mixture-of-experts (MoE) variants, that are SOTA in pre-training metrics and achieve competitive performance after supervised fine-tuning on a range of established multimodal benchmarks. Thanks to large-scale pre-training, MM1 enjoys appealing properties such as enhanced in-context learning, and multi-image reasoning, enabling few-shot chain-of-thought prompting.

cs.CV↗

arXiv2019

Trend-Based Networking Driven by Big Data Telemetry for SDN and Traditional Networks

Organizations face a challenge of accurately analyzing network data and providing automated action based on the observed trend. This trend-based analytics is beneficial to minimize the downtime and improve the performance of the network services, but organizations use different network management tools to understand and visualize the network traffic with limited abilities to dynamically optimize the network. This research focuses on the development of an intelligent system that leverages big data telemetry analysis in Platform for Network Data Analytics (PNDA) to enable comprehensive trend-based networking decisions. The results include a graphical user interface (GUI) done via a web application for effortless management of all subsystems, and the system and application developed in this research demonstrate the true potential for a scalable system capable of effectively benchmarking the network to set the expected behavior for comparison and trend analysis. Moreover, this research provides a proof of concept of how trend analysis results are actioned in both a traditional network and a software-defined network (SDN) to achieve dynamic, automated load balancing.

cs.NI↗

arXiv2011

CBLOCK: An Automatic Blocking Mechanism for Large-Scale De-duplication Tasks

De-duplication---identification of distinct records referring to the same real-world entity---is a well-known challenge in data integration. Since very large datasets prohibit the comparison of every pair of records, {\em blocking} has been identified as a technique of dividing the dataset for pairwise comparisons, thereby trading off {\em recall} of identified duplicates for {\em efficiency}. Traditional de-duplication tasks, while challenging, typically involved a fixed schema such as Census data or medical records. However, with the presence of large, diverse sets of structured data on the web and the need to organize it effectively on content portals, de-duplication systems need to scale in a new dimension to handle a large number of schemas, tasks and data sets, while handling ever larger problem sizes. In addition, when working in a map-reduce framework it is important that canopy formation be implemented as a {\em hash function}, making the canopy design problem more challenging. We present CBLOCK, a system that addresses these challenges. CBLOCK learns hash functions automatically from attribute domains and a labeled dataset consisting of duplicates. Subsequently, CBLOCK expresses blocking functions using a hierarchical tree structure composed of atomic hash functions. The application may guide the automated blocking process based on architectural constraints, such as by specifying a maximum size of each block (based on memory requirements), impose disjointness of blocks (in a grid environment), or specify a particular objective function trading off recall for efficiency. As a post-processing step to automatically generated blocks, CBLOCK {\em rolls-up} smaller blocks to increase recall. We present experimental results on two large-scale de-duplication datasets at Yahoo!---consisting of over 140K movies and 40K restaurants respectively---and demonstrate the utility of CBLOCK.

cs.DB↗

arXiv2001

Mragyati : A System for Keyword-based Searching in Databases

The web, through many search engine sites, has popularized the keyword-based search paradigm, where a user can specify a string of keywords and expect to retrieve relevant documents, possibly ranked by their relevance to the query. Since a lot of information is stored in databases (and not as HTML documents), it is important to provide a similar search paradigm for databases, where users can query a database without knowing the database schema and database query languages such as SQL. In this paper, we propose such a database search system, which accepts a free-form query as a collection of keywords, translates it into queries on the database using the database metadata, and presents query results in a well-structured and browsable form. Th eysytem maps keywords onto the database schema and uses inter-relationships (i.e., data semantics) among the referred tables to generate meaningful query results. We also describe our prototype for database search, called Mragyati. Th eapproach proposed here is scalable, as it does not build an in-memory graph of the entire database for searching for relationships among the objects selected by the user's query.

cs.DB↗