Abstract

Customer support is an essential part of any company to ensure customer satisfaction and business growth. The responsibility and scale get even higher for enterprise companies like Freshworks that build customer support platforms for its enterprise customers from multiple and previously unseen domains. A ticket may carry issues never seen before, therefore static, and predefined problem labels will not always work. These predefined problem labels are often created manually by admins. For a large customer account, roughly 10-35% of the problem labels could get discovered by this manual process and only 5-9% of the available open tickets could get linked to a problem label. So, there is a need for a system for a new incoming ticket. The system should either automatically detect an existing problem topic with its possible suggestions or systematically generate new problem topic(s) for its assignment. Our system is based on an incremental approach to mine ticket data streams and assigns them to hierarchical soft clusters with auto-generated problem/issue topics. Our system achieves an improvement in ticket assignments up to 97%. To compare with recent clustering advancements, we’ve also evaluated our approach on a public dataset [9] and achieved comparable results. Source code and results of the evaluation are available in detail at: https://github.com/vinidixit/hierarchical-labelled-clustering-evaluation.

Abstract

Grain segmentation of sandstone that is partitioning the grain from its surrounding matrix/cement in the thin section is the primary step for computer-aided mineral identification and sandstone classification. The microscopic images of sandstone contain many mineral grains and their surrounding matrix/cement. The distinction between adjacent grains and the matrix is often ambiguous, making grain segmentation difficult. Various solutions exist in literature to handle these problems; however, they are not robust against sandstone petrography's varied pattern. In this paper, we formulate grain segmentation as a pixel-wise two-class (i.e., grain and background) semantic segmentation task. We develop a deep learning-based end-to-end trainable framework named Deep Semantic Grain Segmentation network (DSGSN), a data-driven method, and provide a generic solution. As per the authors' knowledge, this is the first work where the deep neural network is explored to solve the grain segmentation problem. Extensive experiments on microscopic images highlight that our method obtains better segmentation accuracy than various segmentation architectures with more parameters.

Abstract

The goal of this study is to investigate advanced signal processing approaches [single frequency filtering (SFF) and zero-time windowing (ZTW)] with modern deep neural networks (DNNs) [convolution neural networks (CNNs), temporal convolution neural networks (TCN), time-delay neural network (TDNN), and emphasized channel attention, propagation and aggregation in TDNN (ECAPA-TDNN)] for dialect classification of major dialects of English. Previous studies indicated that SFF and ZTW methods provide higher spectro-temporal resolution. To capture the intrinsic variations in articulations among dialects, four feature representations [spectrogram (SPEC), cepstral coefficients, mel filter-bank energies, and mel-frequency cepstral coefficients (MFCCs)] are derived from SFF and ZTW methods. Experiments with and without data augmentation using CNN classifiers revealed that the proposed features performed better than baseline short-time Fourier transform (STFT)-based features on the UT-Podcast database [Hansen, J. H., and Liu, G. (2016). "Unsupervised accent classification for deep data fusion of accent and language information," Speech Commun. 78, 19-33]. Even without data augmentation, all the proposed features showed an approximate improvement of 15%-20% (relative) over best baseline (SPEC-STFT) feature. TCN, TDNN, and ECAPA-TDNN classifiers that capture wider temporal context further improved the performance for many of the proposed and baseline features. Among all the baseline and proposed features, the best performance is achieved with single frequency filtered cepstral coefficients for TCN (81.30%), TDNN (81.53%), and ECAPA-TDNN (85.48%). An investigation of data-driven filters, instead of fixed mel-scale, improved the performance by 2.8% and 1.4% (relatively) for SPEC-STFT and SPEC-SFF, and nearly equal for SPEC-ZTW. To assist related work, we have made the code available ([Kethireddy, R., and Kadiri, S. R. (2022). "Deep neural architectures for dialect classification with single frequency filtering and zero-time windowing feature representations," https://github.com/r39ashmi/e2e_dialect (Last viewed 21 December 2021)].).

Abstract

Measurement error models are studied extensively in the literature. In these models, when the measurement error variance ΣδδΣδδ is known, the estimating techniques require positive definiteness of the matrix Sxx−ΣδδSxx−Σδδ, even when this is positive definite, it might be near singular if the number of observations is small. There are alternative estimators discussed in literature when this matrix is not positive definite. In this paper, estimators when the matrix Sxx−ΣδδSxx−Σδδ is near singular are proposed and it is shown that these estimators are consistent and have the same asymptotic properties as the earlier ones. In addition, we show that our estimators work far better than the earlier estimators in case of small samples and equally good for large samples.

Abstract

The present article deals with the positive solutions of a periodically perturbed Duffing type equation with singularity of the form and its damped counterpart , for . Using a result on monotonicity for the period function of an associated planar autonomous system, we prove the existence of infinitely many subharmonic solutions, as well as the presence of chaotic dynamics, for some T-periodic forcing terms with negative mean value.

Abstract

The information system of healthcare operates through various frameworks, like wireless body area network, telecare medical information system, and mobile or electronic healthcare. All these systems need to maintain the personal health records (PHRs) for various users (ie, patients, doctors, and nurses). In such systems, we need to process and store the health related sensitive data (ie, PHRs). In this article, we aim to provide a robust security mechanism to secure exchange and storage of healthcare data, especially PHRs. We present a generic architecture of blockchain‐enabled secure communication mechanism for Internet of Things‐driven personal health records (BIPHRS). We then discuss various threats and security attacks of healthcare system along with different available security mechanisms. The conducted security analysis and detailed comparative study of the state of art blockchain enabled security schemes for PHR systems show that the proposed BIPHRS provides a better security and more functionality features as compared to other similar existing approaches. We propose a blockchain‐enabled secure communication framework for Internet of Things‐driven personal health records (BIPHRS). The proposed BIPHRS provides better security and more functionality features as compared to the other similar schemes. In addition, a practical

Abstract

Industry 5.0 is the automation, digitization, and data communication of the industrial procedure that comprises industrial cyber–physical systems (I-CPSs), industrial Internet of Things (IIoT), and artificial intelligence (AI). In the I-CPS-enabled healthcare ecosystem, intelligent wearable devices have been extensively employed to sense body information and measure the health status of the patients. Besides other IIoT applications, the I-CPS-enabled healthcare ecosystem also bears various challenges. For instance, due to the communal communication mediums, the security of a patient’s physiological datum is becoming a significant challenge these days. In order to cope with this challenge, we presented a secure and lightweight key establishment protocol. To the best of our knowledge, this protocol is the first application of physically unclonable function (PUF) in the I-CPS-enabled healthcare. The security of the designed protocol is proved with the help of a widely recognized real-or-random (ROR) model. The practical demonstration of our protocol from the network perspective is also measured through broadly recognized NS3 simulator tool.

Abstract

Recent years have witnessed a massive growth in the proliferation of fake news online. User-generated content is a blend of text and visual information leading to producing different variants of fake news. As a result, researchers started targeting multimodal methods for fake news detection. Existing methods capture high-level information from different modalities and jointly model them to decide. Given multiple input modalities, we hypothesize that not all modalities may be equally responsible for decision-making. Hence, this paper presents a novel architecture that effectively identifies and suppresses information from weaker modalities and extracts relevant information from the strong modality on a per-sample basis. We also establish intra-modality relationship by extracting fine-grained image and text features. We conduct extensive experiments on real-world datasets to show that our approach outperforms the state-of-the-art by an average of 3.05% and 4.525% on accuracy and F1-score, respectively. We also release the code, implementation details, and model checkpoints for the community’s interest.1

Abstract

NFT or Non-Fungible Token is a token that certifies a digital asset to be unique. A wide range of assets including, digital art, music, tweets, memes, are being sold as NFTs. NFT-related content has been widely shared on social media sites such as Twitter. We aim to understand the dominant factors that influence NFT asset valuation. Towards this objective, we create a first-of-its-kind dataset linking Twitter and OpenSea (the largest NFT marketplace) to capture social media profiles and linked NFT assets. Our dataset contains 245,159 tweets posted by 17,155 unique users, directly linking 62,997 NFT assets on OpenSea worth 19 Million USD. We have made the dataset public. We analyze the growth of NFTs, characterize the Twitter users promoting NFT assets, and gauge the impact of Twitter features on the virality of an NFT. Further, we investigate the effectiveness of different social media and NFT platform features by experimenting with multiple machine learning and deep learning models to predict an asset's value. Our results show that social media features improve the accuracy by 6% over baseline models that use only NFT platform features. Among social media features, count of user membership lists, number of likes and retweets are important features.

Abstract

Deepfakes are synthetic content generated using advanced deep learning and AI technologies. The advancement of technology has created opportunities for anyone to create and share deepfakes