Searcharxiv⌕ Search

arXiv subjects

Rishabh Jain

Publications and source records attributed to Rishabh Jain.

At least 55 records · Page 3Linked to original sources
arXiv2023

Automatic Concept Embedding Model (ACEM): No train-time concepts, No issue!

Interpretability and explainability of neural networks is continuously increasing in importance, especially within safety-critical domains and to provide the social right to explanation. Concept based explanations align well with how humans reason, proving to be a good way to explain models. Concept Embedding Models (CEMs) are one such concept based explanation architectures. These have shown to overcome the trade-off between explainability and performance. However, they have a key limitation -- they require concept annotations for all their training data. For large datasets, this can be expensive and infeasible. Motivated by this, we propose Automatic Concept Embedding Models (ACEMs), which learn the concept annotations automatically.

cs.LG↗
arXiv2023

Adaptation of Whisper models to child speech recognition

Automatic Speech Recognition (ASR) systems often struggle with transcribing child speech due to the lack of large child speech datasets required to accurately train child-friendly ASR models. However, there are huge amounts of annotated adult speech datasets which were used to create multilingual ASR models, such as Whisper. Our work aims to explore whether such models can be adapted to child speech to improve ASR for children. In addition, we compare Whisper child-adaptations with finetuned self-supervised models, such as wav2vec2. We demonstrate that finetuning Whisper on child speech yields significant improvements in ASR performance on child speech, compared to non finetuned Whisper models. Additionally, utilizing self-supervised Wav2vec2 models that have been finetuned on child speech outperforms Whisper finetuning.

eess.AS↗
arXiv2023

Neural Priority Queues for Graph Neural Networks

Graph Neural Networks (GNNs) have shown considerable success in neural algorithmic reasoning. Many traditional algorithms make use of an explicit memory in the form of a data structure. However, there has been limited exploration on augmenting GNNs with external memory. In this paper, we present Neural Priority Queues, a differentiable analogue to algorithmic priority queues, for GNNs. We propose and motivate a desiderata for memory modules, and show that Neural PQs exhibit the desiderata, and reason about their use with algorithmic reasoning. This is further demonstrated by empirical results on the CLRS-30 dataset. Furthermore, we find the Neural PQs useful in capturing long-range interactions, as empirically shown on a dataset from the Long-Range Graph Benchmark.

cs.LG↗
arXiv2023

FODVid: Flow-guided Object Discovery in Videos

Segmentation of objects in a video is challenging due to the nuances such as motion blurring, parallax, occlusions, changes in illumination, etc. Instead of addressing these nuances separately, we focus on building a generalizable solution that avoids overfitting to the individual intricacies. Such a solution would also help us save enormous resources involved in human annotation of video corpora. To solve Video Object Segmentation (VOS) in an unsupervised setting, we propose a new pipeline (FODVid) based on the idea of guiding segmentation outputs using flow-guided graph-cut and temporal consistency. Basically, we design a segmentation model incorporating intra-frame appearance and flow similarities, and inter-frame temporal continuation of the objects under consideration. We perform an extensive experimental analysis of our straightforward methodology on the standard DAVIS16 video benchmark. Though simple, our approach produces results comparable (within a range of ~2 mIoU) to the existing top approaches in unsupervised VOS. The simplicity and effectiveness of our technique opens up new avenues for research in the video domain.

cs.CV↗
arXiv2023

VGFlow: Visibility guided Flow Network for Human Reposing

The task of human reposing involves generating a realistic image of a person standing in an arbitrary conceivable pose. There are multiple difficulties in generating perceptually accurate images, and existing methods suffer from limitations in preserving texture, maintaining pattern coherence, respecting cloth boundaries, handling occlusions, manipulating skin generation, etc. These difficulties are further exacerbated by the fact that the possible space of pose orientation for humans is large and variable, the nature of clothing items is highly non-rigid, and the diversity in body shape differs largely among the population. To alleviate these difficulties and synthesize perceptually accurate images, we propose VGFlow. Our model uses a visibility-guided flow module to disentangle the flow into visible and invisible parts of the target for simultaneous texture preservation and style manipulation. Furthermore, to tackle distinct body shapes and avoid network artifacts, we also incorporate a self-supervised patch-wise "realness" loss to improve the output. VGFlow achieves state-of-the-art results as observed qualitatively and quantitatively on different image quality metrics (SSIM, LPIPS, FID).

cs.CV↗
arXiv2023

UMFuse: Unified Multi View Fusion for Human Editing applications

Numerous pose-guided human editing methods have been explored by the vision community due to their extensive practical applications. However, most of these methods still use an image-to-image formulation in which a single image is given as input to produce an edited image as output. This objective becomes ill-defined in cases when the target pose differs significantly from the input pose. Existing methods then resort to in-painting or style transfer to handle occlusions and preserve content. In this paper, we explore the utilization of multiple views to minimize the issue of missing information and generate an accurate representation of the underlying human model. To fuse knowledge from multiple viewpoints, we design a multi-view fusion network that takes the pose key points and texture from multiple source images and generates an explainable per-pixel appearance retrieval map. Thereafter, the encodings from a separate network (trained on a single-view human reposing task) are merged in the latent space. This enables us to generate accurate, precise, and visually coherent images for different editing tasks. We show the application of our network on two newly proposed tasks - Multi-view human reposing and Mix&Match Human Image generation. Additionally, we study the limitations of single-view editing and scenarios in which multi-view provides a better alternative.

cs.CV↗
arXiv2023

Parameter Efficient Local Implicit Image Function Network for Face Segmentation

Face parsing is defined as the per-pixel labeling of images containing human faces. The labels are defined to identify key facial regions like eyes, lips, nose, hair, etc. In this work, we make use of the structural consistency of the human face to propose a lightweight face-parsing method using a Local Implicit Function network, FP-LIIF. We propose a simple architecture having a convolutional encoder and a pixel MLP decoder that uses 1/26th number of parameters compared to the state-of-the-art models and yet matches or outperforms state-of-the-art models on multiple datasets, like CelebAMask-HQ and LaPa. We do not use any pretraining, and compared to other works, our network can also generate segmentation at different resolutions without any changes in the input resolution. This work enables the use of facial segmentation on low-compute or low-bandwidth devices because of its higher FPS and smaller model size.

cs.CV↗
arXiv2023

A Wav2vec2-Based Experimental Study on Self-Supervised Learning Methods to Improve Child Speech Recognition

Despite recent advancements in deep learning technologies, Child Speech Recognition remains a challenging task. Current Automatic Speech Recognition (ASR) models require substantial amounts of annotated data for training, which is scarce. In this work, we explore using the ASR model, wav2vec2, with different pretraining and finetuning configurations for self-supervised learning (SSL) toward improving automatic child speech recognition. The pretrained wav2vec2 models were finetuned using different amounts of child speech training data, adult speech data, and a combination of both, to discover the optimum amount of data required to finetune the model for the task of child ASR. Our trained model achieves the best Word Error Rate (WER) of 7.42 on the MyST child speech dataset, 2.99 on the PFSTAR dataset and 12.47 on the CMU KIDS dataset as compared to any other previous methods. Our models outperformed the wav2vec2 BASE 960 on child speech which is considered a state-of-the-art ASR model on adult speech by just using 10 hours of child speech data in finetuning. The analysis of different types of training data and their effect on inference is also provided by using a combination of datasets in pretraining, finetuning and inference.

eess.AS↗
arXiv2023

Extending Logic Explained Networks to Text Classification

Recently, Logic Explained Networks (LENs) have been proposed as explainable-by-design neural models providing logic explanations for their predictions. However, these models have only been applied to vision and tabular data, and they mostly favour the generation of global explanations, while local ones tend to be noisy and verbose. For these reasons, we propose LENp, improving local explanations by perturbing input words, and we test it on text classification. Our results show that (i) LENp provides better local explanations than LIME in terms of sensitivity and faithfulness, and (ii) logic explanations are more useful and user-friendly than feature scoring provided by LIME as attested by a human survey.

cs.CL↗
arXiv2022

Analysis of Distributed Deep Learning in the Cloud

We aim to resolve this problem by introducing a comprehensive distributed deep learning (DDL) profiler, which can determine the various execution "stalls" that DDL suffers from while running on a public cloud. We have implemented the profiler by extending prior work to additionally estimate two types of communication stalls - interconnect and network stalls. We train popular DNN models using the profiler to characterize various AWS GPU instances and list their advantages and shortcomings for users to make an informed decision. We observe that the more expensive GPU instances may not be the most performant for all DNN models and AWS may sub-optimally allocate hardware interconnect resources. Specifically, the intra-machine interconnect can introduce communication overheads up to 90% of DNN training time and network-connected instances can suffer from up to 5x slowdown compared to training on a single instance. Further, we model the impact of DNN macroscopic features such as the number of layers and the number of gradients on communication stalls. Finally, we propose a measurement-based recommendation model for users to lower their public cloud monetary costs for DDL, given a time budget.

cs.LG↗
arXiv2022

Searching for Charged Higgs Bosons via $e^+ e^- \to H^+ H^- \to c\bar{b} \bar{c}b $ at Linear Colliders

We study a search for the charged Higgs boson via $e^+e^- \to H^+H^- \to c\bar{b}\bar{c}b$ at the 500 GeV ILC. In a general two Higgs doublet model without $Z_2$ symmetry, extra Yukawa couplings $ρ_{tt}$ and $ρ_{tc}$ can drive baryogenesis, but searches at the HL-LHC may still go empty-handed if the couplings are relatively weak. Taking $m_{H^+ } \simeq m_H \simeq m_A \simeq 200$ GeV, with $ρ_{tt}$, $ρ_{tc}\sim 0.1$ and no $h(125)$-$H$ mixing, $H^+ \to c\bar b$ decay is dominant, and the $c\bar{b}\bar{c}b$ final state is likely overwhelmed by QCD background at the LHC. We show that the electroweak production of $H^+ H^-$ at the ILC is discoverable with integrated luminosity of 1 ab$^{-1}$. Furthermore, we show that $m_{H^+}$ can be extracted by requiring the two pairs of $b$ and light jets be roughly equal in mass, without assuming the mass value. Thus, ILC can probe low mass Higgs bosons in multijet final states to complement HL-LHC in the future

hep-ph↗
arXiv2022

Detecting Heavy Higgs Bosons from Natural SUSY at a 100 TeV Hadron Collider

Supersymmetric models with radiatively-driven naturalness (RNS) enjoy low electroweak fine-tuning whilst respecting LHC search limits on gluinos and top squarks and allowing for $m_h\simeq 125$ GeV. While the heavier Higgs bosons $H,\ A$ may have TeV-scale masses, the SUSY conserving $μ$ parameter must lie in the few hundred GeV range. Thus, in natural SUSY models there should occur large heavy Higgs boson branching fractions to electroweakinos, with Higgs boson decays to higgsino plus gaugino dominating when they are kinematically accessible. These SUSY decays can open up new avenues for discovery. We investigate the prospects of discovering heavy neutral Higgs bosons $H$ and $A$ decaying into light plus heavy chargino pairs which can yield a four isolated lepton plus missing transverse energy signature at the LHC and at a future 100 TeV $pp$ collider. We find that discovery of heavy Higgs decay to electroweakinos via its $4\ell$ decay mode is very difficult at HL-LHC. For FCC-hh or SPPC, we study the $H,\ A \to $ SUSY reaction along with dominant physics backgrounds from the Standard Model and devise suitable selection requirements to extract a clean signal for FCC-hh or SPPC with $\sqrt{s}=100$ TeV, assuming an integrated luminosity of 15 $ab^{-1}$. We find that while a conventional cut-and-count analysis yields a signal statistical significance greater than $5σ$ for $m_{A,H}\sim 1.1-1.65$ TeV, a boosted-decision-tree analysis allows for heavy Higgs signal discovery at FCC-hh or SPPC for $m_{A,H}\sim 1-2$ TeV.

hep-ph↗
arXiv2022

A Text-to-Speech Pipeline, Evaluation Methodology, and Initial Fine-Tuning Results for Child Speech Synthesis

Speech synthesis has come a long way as current text-to-speech (TTS) models can now generate natural human-sounding speech. However, most of the TTS research focuses on using adult speech data and there has been very limited work done on child speech synthesis. This study developed and validated a training pipeline for fine-tuning state-of-the-art (SOTA) neural TTS models using child speech datasets. This approach adopts a multi-speaker TTS retuning workflow to provide a transfer-learning pipeline. A publicly available child speech dataset was cleaned to provide a smaller subset of approximately 19 hours, which formed the basis of our fine-tuning experiments. Both subjective and objective evaluations were performed using a pretrained MOSNet for objective evaluation and a novel subjective framework for mean opinion score (MOS) evaluations. Subjective evaluations achieved the MOS of 3.95 for speech intelligibility, 3.89 for voice naturalness, and 3.96 for voice consistency. Objective evaluation using a pretrained MOSNet showed a strong correlation between real and synthetic child voices. Speaker similarity was also verified by calculating the cosine similarity between the embeddings of utterances. An automatic speech recognition (ASR) model is also used to provide a word error rate (WER) comparison between the real and synthetic child voices. The final trained TTS model was able to synthesize child-like speech from reference audio samples as short as 5 seconds.

cs.SD↗
arXiv2022

On Hybrid Quantum and Classical Computing Algorithms for Mixed-Integer Programming

Quantum computing is emerging as a new computing resource that could be superior to conventional computing for certain classes of optimization problems. However, in principle, most existing approaches to quantum optimization are intended to solve unconstrained binary programming problems, while mixed-integer linear programming is of most interest in practice. We attempt to bridge the gap between the capability of quantum computing and real-world applications by developing a new approach for mixed-integer programming. The approach applies Benders decomposition to decompose the mixed-integer programming into binary programming and linear programming sub-problems, which are solved by a noisy intermediate-scale quantum processor and conventional processor, respectively. The algorithm is provably able to reach the optimal solution of the original mixed-integer programming problem. The algorithm is tested on a D-Wave 2000Q quantum processing unit and is shown to be effective for small-scaled test cases. We also test the algorithm on a mixed-integer programming inspired by power system applications. Many insights are drawn from the numerical results for both the capabilities and limitations of the proposed algorithm.

math.OC↗
arXiv2021

Collider Prospects for Muon $g-2$ in General Two Higgs Doublet Model

Recent progress on muon $g-2$ measurement prompts one to take it even more seriously. In the general two Higgs doublet model that allows extra Yukawa couplings, we take a simplified approach of single enhanced coupling. We fix the charged lepton flavor violating coupling, $ρ_{τμ} = ρ_{μτ}$, via the one-loop mechanism, for illustrative masses of the heavy scalar $H$ and pseudoscalar $A$, where we assume $m_A = m_{H^+}$. Since extra top Yukawa couplings are plausibly the largest, we turn on $ρ_{tt}$ and find that LHC search for $gg \to H,\,A \to τμ$ gives more stringent bound than from $τ\to μγ$ with two-loop mechanism. Turning on a second extra top Yukawa coupling, $ρ_{tc}$, can loosen the bound on $ρ_{tt}$, but LHC constraints can again be more stringent than from $B \to Dμν$ vs $Deν$ universality. This means that evidence for $H,\,A \to τμ$ may yet emerge with full LHC Run 2 data, while direct search for $τ^\pmμ^\mp bW^+$ or $t\bar cbW^+$ (plus conjugate) may also bear fruit.

hep-ph↗
arXiv2021

ZFlow: Gated Appearance Flow-based Virtual Try-on with 3D Priors

Image-based virtual try-on involves synthesizing perceptually convincing images of a model wearing a particular garment and has garnered significant research interest due to its immense practical applicability. Recent methods involve a two stage process: i) warping of the garment to align with the model ii) texture fusion of the warped garment and target model to generate the try-on output. Issues arise due to the non-rigid nature of garments and the lack of geometric information about the model or the garment. It often results in improper rendering of granular details. We propose ZFlow, an end-to-end framework, which seeks to alleviate these concerns regarding geometric and textural integrity (such as pose, depth-ordering, skin and neckline reproduction) through a combination of gated aggregation of hierarchical flow estimates termed Gated Appearance Flow, and dense structural priors at various stage of the network. ZFlow achieves state-of-the-art results as observed qualitatively, and on quantitative benchmarks of image quality (PSNR, SSIM, and FID). The paper presents extensive comparisons with other existing solutions including a detailed user study and ablation studies to gauge the effect of each of our contributions on multiple datasets.

cs.CV↗
arXiv2021

Flavor changing top decays to charm and Higgs with $ττ$ at the LHC

We investigate the prospects of discovering the top quark decay into a charm quark and a Higgs boson ($t \to c h^0$) in top quark pair production at the CERN Large Hadron Collider (LHC). A general two Higgs doublet model is adopted to study flavor changing neutral Higgs (FCNH) interactions. We perform a parton level analysis as well as Monte Carlo simulations using \textsc{Pythia}~8 and \textsc{Delphes} to study the flavor changing top quark decay $t \to c h^0$, followed by the Higgs decaying into $τ^+ τ^-$, with the other top quark decaying to a bottom quark ($b$) and two light jets ($t\to bW\to bjj$). To reduce the physics background to the Higgs signal, only the leptonic decays of tau leptons are used, $τ^+τ^- \to e^\pmμ^\mp +\slashed{E}_T$, where $\slashed{E}_T$ represents the missing transverse energy from the neutrinos. In order to reconstruct the Higgs boson and top quark masses as well as to effectively remove the physics background, the collinear approximation for the highly boosted tau decays is employed. Our analysis suggests that a high energy LHC at $\sqrt{s} = 27$ TeV will be able to discover this FCNH signal with an integrated luminosity $\mathcal{L} = 3$ ab$^{-1}$ for a branching fraction ${\cal B}(t \to ch^0) \agt 1.4 \times 10^{-4}$ that corresponds to a FCNH coupling $|λ_{tch}| \agt 0.023$. This FCNH coupling is significantly below the current ATLAS combined upper limit of $|λ_{tch}| = 0.064$.

hep-ph↗
arXiv2020

A Search for Technosignatures Around 31 Sun-like Stars with the Green Bank Telescope at 1.15-1.73 GHz

We conducted a search for technosignatures in April of 2018 and 2019 with the L-band receiver (1.15-1.73 GHz) of the 100 m diameter Green Bank Telescope. These observations focused on regions surrounding 31 Sun-like stars near the plane of the Galaxy. We present the results of our search for narrowband signals in this data set as well as improvements to our data processing pipeline. Specifically, we applied an improved candidate signal detection procedure that relies on the topographic prominence of the signal power, which nearly doubles the signal detection count of some previously analyzed data sets. We also improved the direction-of-origin filters that remove most radio frequency interference (RFI) to ensure that they uniquely link signals observed in separate scans. We performed a preliminary signal injection and recovery analysis to test the performance of our pipeline. We found that our pipeline recovers 93% of the injected signals over the usable frequency range of the receiver and 98% if we exclude regions with dense RFI. In this analysis, 99.73% of the recovered signals were correctly classified as technosignature candidates. Our improved data processing pipeline classified over 99.84% of the ~26 million signals detected in our data as RFI. Of the remaining candidates, 4539 were detected outside of known RFI frequency regions. The remaining candidates were visually inspected and verified to be of anthropogenic nature. Our search compares favorably to other recent searches in terms of end-to-end sensitivity, frequency drift rate coverage, and signal detection count per unit bandwidth per unit integration time.

astro-ph.EP↗