arXiv subjects

Anand Kulkarni

Publications and source records attributed to Anand Kulkarni.

5 recordsLinked to original sources

arXiv2023

Construction of Multi-Dimensional Functions for Optimization of Additive-Manufacturing Process Parameters

The authors present a generic framework for the parameter optimization of additive manufacturing (AM) processes, one tailored to a high-throughput experimental methodology (HTEM). Given the large number of parameters, which impact the quality of AM-metallic components, the authors advocate for partitioning the AM parameter set into stages (tiers), based on their relative importance, modeling one tier at a time until successful, and then systematically expanding the framework. The authors demonstrate how the construction of multi-dimensional functions, based on neural networks (NN), can be applied to successfully model relative densities and Rockwell hardness obtained from HTEM testing of the Inconel 718 superalloy fabricated, using a powder-bed approach. The authors analyze the input data set, assess its suitability for predictions, and show how to optimize the framework for the multi-dimensional functional construction, such as to obtain the highest degree of fit with the input data set. The authors also compare and contrast the NN-based multi-dimensional functional construction to multi-variate linear regression, to polynomial regression, and to Gaussian process regression (GPR), highlight similarity between the NN-based multi-dimensional functional construction and the GPR, and offer insights into the suitability of each of these methods for the data set and the application at hand. In terms of the coefficient of determination, $R^2$, a relatively simple, single-layer NN with 5 or 10 nodes outperforms multi-variate linear regression, 2nd-order polynomial regression, and GPR for the primary Inconel 718 HTEM data set studied. The novelty of the research work entails the versatile and scalable NN framework presented, suitable for use in conjunction with HTEM, for the AM parameter optimization of superalloys, and beyond.

cond-mat.mtrl-sci↗

arXiv2021

MAVE: A Product Dataset for Multi-source Attribute Value Extraction

Attribute value extraction refers to the task of identifying values of an attribute of interest from product information. Product attribute values are essential in many e-commerce scenarios, such as customer service robots, product ranking, retrieval and recommendations. While in the real world, the attribute values of a product are usually incomplete and vary over time, which greatly hinders the practical applications. In this paper, we introduce MAVE, a new dataset to better facilitate research on product attribute value extraction. MAVE is composed of a curated set of 2.2 million products from Amazon pages, with 3 million attribute-value annotations across 1257 unique categories. MAVE has four main and unique advantages: First, MAVE is the largest product attribute value extraction dataset by the number of attribute-value examples. Second, MAVE includes multi-source representations from the product, which captures the full product information with high attribute coverage. Third, MAVE represents a more diverse set of attributes and values relative to what previous datasets cover. Lastly, MAVE provides a very challenging zero-shot test set, as we empirically illustrate in the experiments. We further propose a novel approach that effectively extracts the attribute value from the multi-source product information. We conduct extensive experiments with several baselines and show that MAVE is an effective dataset for attribute value extraction task. It is also a very challenging task on zero-shot attribute extraction. Data is available at {\it \url{https://github.com/google-research-datasets/MAVE}}.

cs.CL↗

arXiv2020

Machine Learning and Data Analytics for Design and Manufacturing of High-Entropy Materials Exhibiting Mechanical or Fatigue Properties of Interest

This chapter presents an innovative framework for the application of machine learning and data analytics for the identification of alloys or composites exhibiting certain desired properties of interest. The main focus is on alloys and composites with large composition spaces for structural materials. Such alloys or composites are referred to as high-entropy materials (HEMs) and are here presented primarily in context of structural applications. For each output property of interest, the corresponding driving (input) factors are identified. These input factors may include the material composition, heat treatment, manufacturing process, microstructure, temperature, strain rate, environment, or testing mode. The framework assumes the selection of an optimization technique suitable for the application at hand and the data available. Physics-based models are presented, such as for predicting the ultimate tensile strength (UTS) or fatigue resistance. We devise models capable of accounting for physics-based dependencies. We factor such dependencies into the models as a priori information. In case that an artificial neural network (ANN) is deemed suitable for the applications at hand, it is suggested to employ custom kernel functions consistent with the underlying physics, for the purpose of attaining tighter coupling, better prediction, and for extracting the most out of the - usually limited - input data available.

cond-mat.mtrl-sci↗

arXiv2014

Exchanging Ohmic Losses in Metamaterial Absorbers with Useful Optical Absorption for Photovoltaics

Using metamaterial absorbers, we have shown that metallic layers in the absorbers do not necessarily constitute undesired resistive heating problem for photovoltaics. Tailoring the geometric skin depth of metals and employing the natural bulk absorbance characteristics of the semiconductors in those absorbers can enable the exchange of undesired resistive losses with the useful optical absorbance in the active semiconductors. Thus, Ohmic loss dominated metamaterial absorbers can be converted into photovoltaic near-perfect absorbers with the advantage of harvesting the full potential of light management offered by the metamaterial absorbers. Based on experimental permittivity data for indium gallium nitride, we have shown that between 75%-95% absorbance can be achieved in the semiconductor layers of the converted metamaterial absorbers. Besides other metamaterial and plasmonic devices, our results may also apply to photodectors and other metal or semiconductor based optical devices where resistive losses and power consumption are important pertaining to the device performance.

physics.optics↗

arXiv2009

Combinatorial Polytope Enumeration

We describe a provably complete algorithm for the generation of a tight, possibly exact superset of all combinatorially distinct simple n-facet polytopes in R^d, along with their graphs, f-vectors, and face lattices. The technique applies repeated cutting planes and planar sweeps to a d-simplex. Our generator has implications for several outstanding problems in polytope theory, including conjectures about the number of distinct polytopes, the edge expansion of polytopal graphs, and the d-step conjecture.

math.CO↗