Source author record

Eunjung Han

Eunjung Han appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

eess.AS Sound Computation and Language Genomics Machine Learning

Catalog footprint

What is connected

4works

5topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

ASR-Aware End-to-end Neural Diarization

We present a Conformer-based end-to-end neural diarization (EEND) model that uses both acoustic input and features derived from an automatic speech recognition (ASR) model. Two categories of features are explored: features derived directly from ASR output (phones, position-in-word and word boundaries) and features derived from a lexical speaker change detection model, trained by fine-tuning a pretrained BERT model on the ASR output. Three modifications to the Conformer-based EEND architecture are proposed to incorporate the features. First, ASR features are concatenated with acoustic features. Second, we propose a new attention mechanism called contextualized self-attention that utilizes ASR features to build robust speaker representations. Finally, multi-task learning is used to train the model to minimize classification loss for the ASR features along with diarization loss. Experiments on the two-speaker English conversations of Switchboard+SRE data sets show that multi-task learning with position-in-word information is the most effective way of utilizing ASR features, reducing the diarization error rate (DER) by 20% relative to the baseline.

preprint2022arXiv

Contrastive-mixup learning for improved speaker verification

This paper proposes a novel formulation of prototypical loss with mixup for speaker verification. Mixup is a simple yet efficient data augmentation technique that fabricates a weighted combination of random data point and label pairs for deep neural network training. Mixup has attracted increasing attention due to its ability to improve robustness and generalization of deep neural networks. Although mixup has shown success in diverse domains, most applications have centered around closed-set classification tasks. In this work, we propose contrastive-mixup, a novel augmentation strategy that learns distinguishing representations based on a distance metric. During training, mixup operations generate convex interpolations of both inputs and virtual labels. Moreover, we have reformulated the prototypical loss function such that mixup is enabled on metric learning objectives. To demonstrate its generalization given limited training data, we conduct experiments by varying the number of available utterances from each speaker in the VoxCeleb database. Experimental results show that applying contrastive-mixup outperforms the existing baseline, reducing error rate by 16% relatively, especially when the number of training utterances per speaker is limited.

preprint2021arXiv

BW-EDA-EEND: Streaming End-to-End Neural Speaker Diarization for a Variable Number of Speakers

We present a novel online end-to-end neural diarization system, BW-EDA-EEND, that processes data incrementally for a variable number of speakers. The system is based on the Encoder-Decoder-Attractor (EDA) architecture of Horiguchi et al., but utilizes the incremental Transformer encoder, attending only to its left contexts and using block-level recurrence in the hidden states to carry information from block to block, making the algorithm complexity linear in time. We propose two variants: For unlimited-latency BW-EDA-EEND, which processes inputs in linear time, we show only moderate degradation for up to two speakers using a context size of 10 seconds compared to offline EDA-EEND. With more than two speakers, the accuracy gap between online and offline grows, but the algorithm still outperforms a baseline offline clustering diarization system for one to four speakers with unlimited context size, and shows comparable accuracy with context size of 10 seconds. For limited-latency BW-EDA-EEND, which produces diarization outputs block-by-block as audio arrives, we show accuracy comparable to the offline clustering-based system.

preprint2013arXiv

Genome Sequencing Highlights Genes Under Selection and the Dynamic Early History of Dogs

To identify genetic changes underlying dog domestication and reconstruct their early evolutionary history, we analyzed novel high-quality genome sequences of three gray wolves, one from each of three putative centers of dog domestication, two ancient dog lineages (Basenji and Dingo) and a golden jackal as an outgroup. We find dogs and wolves diverged through a dynamic process involving population bottlenecks in both lineages and post-divergence gene flow, which confounds previous inferences of dog origins. In dogs, the domestication bottleneck was severe involving a 17 to 49-fold reduction in population size, a much stronger bottleneck than estimated previously from less intensive sequencing efforts. A sharp bottleneck in wolves occurred soon after their divergence from dogs, implying that the pool of diversity from which dogs arose was far larger than represented by modern wolf populations. Conditional on mutation rate, we narrow the plausible range for the date of initial dog domestication to an interval from 11 to 16 thousand years ago. This period predates the rise of agriculture, implying that the earliest dogs arose alongside hunter-gathers rather than agriculturists. Regarding the geographic origin of dogs, we find that surprisingly, none of the extant wolf lineages from putative domestication centers are more closely related to dogs, and the sampled wolves instead form a sister monophyletic clade. This result, in combination with our finding of dog-wolf admixture during the process of domestication, suggests a re-evaluation of past hypotheses of dog origin is necessary. Finally, we also detect signatures of selection, including evidence for selection on genes implicated in morphology, metabolism, and neural development. Uniquely, we find support for selective sweeps at regulatory sites suggesting gene regulatory changes played a critical role in dog domestication.