Improved phone classification using semi-supervised learning with autoencoders.
problem Phone classification accuracy with limited labeled data.
method Semi-supervised learning with sparse autoencoders, using both labeled and unlabelled data.
result The method outperforms standard supervised training and provides competitive error rates.
Optimizes phonetic recognition system to reduce latency.
problem Reducing latency in CD-DNN based phoneme recognition.
method Shifted input window with up to 5 frames of past context.
result Improves latency by 50 ms with asymmetric window.
An important step in speaker verification is extracting features that best characterize the speaker voice. This paper investigates a front-end processing that aims at improving the performance of speaker verification based on the SVMs classifier, in text independent mode. This approach combines features based on conven…
New neural network design improves speech enhancement metrics.
problem Improving speech enhancement metrics in noisy conditions.
method Combination of convolutional and recurrent layers in U-net architecture.
result Proposed solution outperforms current state-of-the-art in SDR, SIR, and STOI metrics.
Improved speech segment clustering using FTDTW.
problem Clustering speech segments of varying lengths and features.
method Feature Trajectory Dynamic Time Warping (FTDTW) for pairwise alignment of feature trajectories.
result Consistent and statistically significant improvements in cluster quality (F-measure and NMI).
End-to-end transformer model improves lexical stress detection accuracy.
problem Inaccurate phoneme boundaries and limited features for stress classification.
method End-to-end sequence to sequence model using transformer trained on feature sequences and phoneme sequences with stress marks.
result End-to-end model achieves better performance and lower phoneme error rate (6.36%) compared to syllable segmentation methods.
End-to-end model detects articulatory features from speech data.
problem Detecting articulatory features from speech data for various applications.
method Apply Listen, Attend and Spell (LAS) architecture and attention models.
result End-to-end training of manners and places of articulation detectors.
Improved phone classification accuracy using graph-based regularization.
problem Phone classification with limited labeled data.
method Graph-based semi-supervised learning with stochastic entropic regularization.
result Significantly improved phone classification accuracy with low labeled data.
Improved speech recognition with audio-visual fusion.
problem Enhance speech recognition accuracy in noisy conditions.
method Proposes an attention-based audio-visual fusion strategy to align and learn from acoustic and lip motion data.
result Significant improvements in recognition accuracy (7-30%) on TCD-TIMIT dataset.
This paper improves speech recognition by distilling knowledge from acoustic models.
problem Improving speech recognition accuracy using ensemble models.
method Proposes multi-teacher distillation strategies for joint CTC-attention end-to-end ASR systems, integrating error rate metric for optimization.
result Reports state-of-the-art error rates on various datasets and languages.
Enhances speech quality in noisy environments using symbolic sequential modeling.
problem Improving speech quality in noisy conditions.
method Incorporates symbolic sequential modeling into speech enhancement framework.
result Significant improvement in speech quality metrics (PESQ, STOI) on TIMIT dataset.
Quaternion neural networks improve speech recognition efficiency.
problem Improving speech recognition systems with fewer parameters.
method Investigating quaternion-valued convolutional and recurrent neural networks.
result Quaternion neural networks outperform real-valued models with fewer parameters.
Unsupervised speech recognition without labeled data using novel cost function and MAP refinement.
problem Training speech recognition systems without labeled data.
method Alternates between phoneme classifier learning and boundary refinement using Segmental Empirical Output Distribution Matching and MAP approach.
result Achieves phone error rate (PER) of 41.6% on TIMIT dataset.
Prob-PIT improves speech separation by considering output-label permutations as random variables.
problem Overconfident output-label assignment in PIT leads to unreliable speech separation.
method Prob-PIT treats output-label permutations as a discrete latent random variable with a uniform prior distribution and maximizes the log-likelihood function.
result Prob-PIT significantly outperforms PIT in terms of Signal to Distortion Ratio and Signal to Interference Ratio.
Aligns databases with Gaussian features using MAP estimation and thresholding.
problem Aligning databases with jointly Gaussian features.
method Two algorithms: MAP estimation and thresholding of log likelihood ratios.
result Conditions on mutual information for reliable performance of algorithms.
We replace the Hidden Markov Model (HMM) which is traditionally used in in continuous speech recognition with a bi-directional recurrent neural network encoder coupled to a recurrent neural network decoder that directly emits a stream of phonemes. The alignment between the input and output sequences is established usin…
Study tackles database variability in medical data using ensemble models and CNNs.
problem Achieving robust generalization in machine learning models across multiple medical databases.
method Ensemble of local models based on convolutional neural networks (CNNs) and various data preprocessing techniques.
result Improved inter-database generalization performance and scalability of models.
Database of Irish parliamentary speeches from 1919-2013.
problem Storing and analyzing historical parliamentary debates.
method Downloaded speeches from official website, processed, parsed, and matched with TD and minister data.
result 4.5 million speeches from 1,178 TDs in a single database.
Study connects database alignment and planted matching using Gaussian features.
problem Identify matching between correlated user features in anonymized databases.
method Derived results for database alignment and planted matching, showing connections and thresholds.
result Performance thresholds for database alignment converge to planted matching when feature dimensionality is sufficiently high.
We propose a distributed approach to train deep neural networks (DNNs), which has guaranteed convergence theoretically and great scalability empirically: close to 6 times faster on instance of ImageNet data set when run with 6 machines. The proposed scheme is close to optimally scalable in terms of number of machines, …
This work applies SQL to deep learning, leveraging database techniques.
problem Applying deep learning techniques to databases.
method Expressing deep learning operations using SQL, a multidimensional array language.
result Demonstrates the feasibility of using SQL for deep learning operations.
In this paper, we propose and study random maxout features, which are constructed by first projecting the input data onto sets of randomly generated vectors with Gaussian elements, and then outputing the maximum projection value for each set. We show that the resulting random feature map, when used in conjunction with …
We stabilize the activations of Recurrent Neural Networks (RNNs) by penalizing the squared distance between successive hidden states' norms. This penalty term is an effective regularizer for RNNs including LSTMs and IRNNs, improving performance on character-level language modeling and phoneme recognition, and outperfor…
Study compares CDS databases and finds discrepancies due to various factors.
problem Comparing discrepancies among CDS databases.
method Comparing five major sources of corporate CDS prices over 2004-2010.
result CMA quotes lead price discovery and databases disagree on stock-CDS return analysis.
Graph database outperforms in filtering ESG stocks efficiently.
problem Efficiently filtering ESG stocks from large lists of equities.
method Compared SQL, No-SQL, and graph databases; used Python for database interactions; collected data from stock price and financial news.
result Graph database is more efficient for ESG stock filtering.
CwA optimizes search performance by jointly learning a balanced database partition and a neural probing function.
problem Suboptimal search performance due to mismatched database and query distributions.
method CwA jointly learns a balanced database partition and a neural probing function to optimize search performance directly for the query distribution.
result CwA achieves up to 4.7x throughput over state-of-the-art methods at equal recall.
PyODDS is a Python system for outlier detection in databases.
problem Outlier detection in databases for various users.
method End-to-end Python system with database support, including statistical and deep learning algorithms.
result Supports outlier detection algorithms for users with or without machine learning background.
Taxonomies of cryptocurrencies and comparisons with fiat money and databases.
problem Comparing cryptocurrencies, fiat money, and databases.
method Taxonomy creation and comparison analysis.
result Cryptocurrencies and databases have unique characteristics not found in fiat money.
Bayesian entity resolution merges together multiple, noisy databases and returns the minimal collection of unique individuals represented, together with their true, latent record values. Bayesian methods allow flexible generative models that share power across databases as well as principled quantification of uncertain…
Bayesian algorithm detects image matches and fraud.
problem Detecting identity matches and fraud in image databases.
method Generative model of image graph trained with matching algorithm.
result Bayesian approach improves detection accuracy.
Paper improves text-to-SQL translation by encoding schema relations with self-attention.
problem Improving text-to-SQL translation accuracy across diverse databases.
method Uses relation-aware self-attention to encode schema information.
result Significant gains on Spider dataset (42.94% exact match accuracy).
Graph Neural Networks improve machine learning on relational databases.
problem Training machine learning models on relational databases requires costly data extraction and feature engineering.
method Uses Graph Neural Networks to extract features from relational databases.
result Outperforms state-of-the-art automatic feature engineering methods.
Paper presents a new Wi-Fi RSS and geomagnetic field database for indoor localization and trajectory estimation.
problem Indoor localization and trajectory estimation challenges.
method Convolutional neural network (CNN) for RSS data and LSTM network for geomagnetic field intensity.
result CNN and LSTM networks show feasibility for localization and trajectory estimation.
New deep learning method validated across multiple sleep staging databases.
problem Improving automatic sleep scoring accuracy across different datasets.
method Ensemble of local models using deep learning for automatic sleep staging.
result Good general performance compared to human experts and state-of-the-art methods.
EERN uses deep learning for relational databases, outperforming other methods.
problem Deep learning for relational databases.
method Equivariant Entity-Relationship Network (EERN) using MLP equivariant to Entity-Relationship model symmetries.
result EERN outperforms other methods in synthetic and real-data experiments.
Super Learner improves prediction in electronic healthcare databases.
problem Improving prediction models in large electronic healthcare databases.
method Super Learner ensemble learning algorithm combined with high-dimensional propensity score variable selection.
result Super Learner adapts to different datasets and optimizes predictive performance.
New method protects privacy while allowing accurate statistical inference from synthetic data.
problem Ensuring privacy in database release while maintaining statistical utility.
method Kernel mean embedding with differential privacy constraints.
result Consistent estimators of population statistics can be constructed while protecting individual privacy.
Extends Aff-Wild database for affect recognition in real-world settings.
problem Complex human emotional states in real-world settings.
method Developed deep neural architectures with attention mechanism for emotion recognition.
result Improved performance in emotion recognition using Aff-Wild2.
Transfer learning improves handwriting recognition in small datasets.
problem Reduced training datasets for handwriting text recognition.
method Transfer learning from a larger database to a smaller one, focusing on retraining the whole CNNLSTM-CTC model.
result Significant improvement in character error rate (CER) with transfer learning, reducing CER from 18.2% to 3.3%.
New vector quantization method reduces relevance of parallel components in database points.
problem Scaling maximum inner product search to massive databases.
method Developed anisotropic vector quantization loss functions.
result Achieves state-of-the-art results on public benchmarks.
Stochastic neural networks improve speech and music modeling.
problem Efficiently propagate uncertainty in latent state representations.
method Introduce stochastic recurrent neural networks that combine deterministic and state space models.
result Significant improvement on Blizzard and TIMIT speech modeling data sets.
We study the distribution of fluctuations over a time scale Δt (i.e., the returns) of the S&P 500 index by analyzing three distinct databases. Database (i) contains approximately 1 million records sampled at 1 min intervals for the 13-year period 1984-1996, database (ii) contains 8686 daily records for the 35-year pe…
Topology-based information retrieval improves query accuracy.
problem Query accuracy in databases with complex structures.
method Dilation-invariant comparative measures of persistent homology.
result Topology-based retrieval outperforms standard methods.
A new algorithm uses bandits to diversify database activity monitoring.
problem Limitation of current DAM systems in collecting diverse data.
method Redefined DAM sampling as a bandit problem and developed a novel algorithm combining expert knowledge and random exploration.
result Adding diversity to sampling using the bandit-based approach improves coverage without decreasing alert quality.
The paper studies causal effects of multiple treatments in healthcare databases with rare outcomes.
problem Estimating causal effects of multiple treatments in healthcare databases with rare outcomes.
method The paper designs three sets of simulations and compares the operating characteristics of three types of methods: Bayesian Additive Regression Trees (BART), regression adjustment on multivariate spline of generalized propensity scores (RAMS), and inverse probability of treatment weighting (IPTW) with multinomial logistic regression or generalized boosted models.
result BART and RAMS provide lower bias and mean squared error compared to IPTW methods.
PyODDS automates outlier detection for new data sources.
problem Manual outlier detection is inefficient and domain-specific.
method Automated end-to-end system with Apache Spark and database support.
result PyODDS optimizes outlier detection pipelines automatically.
Learning invariant representations is an important problem in machine learning and pattern recognition. In this paper, we present a novel framework of transformation-invariant feature learning by incorporating linear transformations into the feature learning algorithms. For example, we present the transformation-invari…
New method embeds DNA sequences for faster, more informative gene comparison.
problem Slow and costly sequence comparison methods for genes without exact matches.
method Recurrent neural networks to embed sequences in a low-dimensional space.
result Embedding allows for better comparison of genes without exact matches.