Auto-encoder model simplifies industrial sensor data into fixed-length vectors.
problem Efficiently summarizing large-scale industrial sensor signals.
method Recurrent auto-encoder with partial reconstruction and rolling window approach.
result Fixed-length vectors capture selected features over time.
Proposes a new model to predict click-through rates by learning user interests.
problem Difficulty in capturing diverse user interests from historical behaviors.
method Introduces a local activation unit to adaptively learn user interests from historical behaviors for each ad.
result Improves model's expressive ability by varying representation vectors over different ads.
RATQ is a new quantizer for optimizing noisy gradients in machine learning.
problem Optimizing noisy gradients in stochastic optimization.
method RATQ uses Hadamard transform and adaptive uniform quantization, and achieves near-optimal performance.
result RATQ nearly achieves information theoretic lower bounds for optimization accuracy.
Rational kernels offer a way to handle sequence data efficiently.
problem Handling sequence data with fixed-length feature vectors is cumbersome.
method Rational kernels provide an elegant representation for sequences.
result Rational kernels offer a convenient way to use sequence data.
A neural network approach solves multiple-instance problems more effectively.
problem Difficult to describe objects by a single vector, requiring bag-level classification.
method Proposes a neural network formalism for MIL problems, optimizing with modified back-propagation.
result Shows superior performance compared to existing classifiers on 14 benchmark datasets.
Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural ma…
The study finds solutions to a curvature minimisation problem in fixed-length curves.
problem Minimizing the L∞-norm of curvature among curves of fixed length. method Characterized solutions by a system of differential equations and classified the structure of solutions.
result Characterized solutions to the L∞-norm of curvature problem. Bayesian approach improves paragraph vector entropy and uncertainty in text analysis.
problem Capturing semantic relationships in text of varying lengths.
method Probabilistic generative model for paragraph vectors with Bayesian inference.
result Entropy of paragraph vectors decreases with document length and uncertainty improves performance in text analysis.
Complex embeddings handle non-metric proximity data better than traditional methods.
problem Proximities not always metric or inner product-based, causing convergence issues.
method Proposes complex-valued embeddings for non-vectorial data.
result Complex embeddings outperform traditional techniques on benchmarks.
A new method classifies heart sounds using i-vectors and machine learning.
problem Heart sound classification for disease diagnostics.
method Extract i-vectors from MFCC features, apply PCA and VAE for dimensionality reduction, then use GMMs and SVM for classification.
result The method improves heart sound classification by 16% on the Physionet dataset.
GraphKKE learns fixed-length feature vectors from time-evolving graphs of human microbiome data.
problem Understanding dynamic changes in human microbiome graphs over time.
method Spectral analysis of transfer operators and graph kernels.
result GraphKKE captures temporal changes in human microbiome graphs.
Convolutional residual model predicts diagnoses from EHR notes.
problem Predicting multiple diagnoses from medical text data.
method Convolutional Neural Network (CNN) + Deep Residual Network.
result Superior performance compared to baseline models.
Model converts code snippets into vectors for predicting method names.
problem Representing code as vectors for semantic analysis.
method Decomposes code into abstract syntax tree paths, learns atomic representations simultaneously with aggregation.
result Code vectors trained on 14M methods can predict method names from unseen files.
NBF combines deep learning with classical filtering for better belief tracking.
problem Maintaining distributions over hidden states in partially observable systems.
method Trains neural networks to map beliefs to fixed-length vectors, updating them with incoming observations and dynamics.
result NBF efficiently tracks shifting, multimodal beliefs without particle impoverishment.
Transformer-XL extends language models' context length without disrupting temporal coherence.
problem Fixed-length context limitation in Transformers.
method Segment-level recurrence and novel positional encoding.
result 80% longer dependency, 450% longer than vanilla Transformers.
In this paper, we propose a novel neural network model called RNN Encoder-Decoder that consists of two recurrent neural networks (RNN). One RNN encodes a sequence of symbols into a fixed-length vector representation, and the other decodes the representation into another sequence of symbols. The encoder and decoder of t…
Analyticity of critical points for O'Hara's knot energies proved.
problem Analyzing the regularity of critical points for O'Hara's knot energies.
method Cauchy's method of majorants and a Möbius energy-inspired gradient decomposition.
result Smooth critical points of O'Hara's knot energies are analytic.
Introduces BWMD, a new distance measure for DNA and malware clustering.
problem Shortcomings of previous compression-based distance metrics.
method Embeds sequences into a fixed-length feature vector.
result Significantly improved clustering performance on larger malware corpora.
We consider the expected value for the total curvature of a random closed polygon. Numerical experiments have suggested that as the number of edges becomes large, the difference between the expected total curvature of a random closed polygon and a random open polygon with the same number of turning angles approaches a …
SummerTime summarizes variable-length time series for machine learning applications.
problem Classical machine learning methods struggle with variable-length time series data.
method Summarizes time series into a fixed-length feature vector using Gaussian Mixture Models (GMM).
result Improves classification and regression performance in physical activity analysis.
Efficiently identifies users from walking activity data using kernel-based DTW.
problem Identifying users from walking activity data streams.
method Learning a kernel to approximate DTW for efficient analysis of streaming data from wearable sensors.
result The proposed approach reduces computational burden compared to traditional DTW.
Examines challenges and proposes new approaches in machine learning theory.
problem Challenges in machine learning as a function approximation and optimization.
method Mathematical analysis of gradient descent, fixed network limitations, and RNNs.
result New insights and mathematical approaches to improve machine learning.
Physical knots and links are one-dimensional submanifolds of R^3 with fixed length and thickness. We show that isotopy classes in this category can differ from those of classical knot and link theory. In particular we exhibit a Gordian Split Link, a two component link that is split in the classical theory but cannot be…
Paper proposes using LSTM for LSH-based sequence alignment.
problem Sequence alignment using deep learning models.
method Deep bidirectional LSTM for feature learning and LSH-based sequence alignment.
result Higher accuracy achieved with LSTM-based model.
Unified pipeline classifies time series using complex networks and persistent homology.
problem Classifying univariate time series using various graph constructions and metrics.
method Time series to graph, graph to dissimilarity matrix, filtration to persistence diagrams, vectorization to features.
result Persistence-based features are robust to noise and optimal graph type depends on signal structure.
Survey of word embedding techniques for NLP.
problem Creating effective word representations for natural language processing.
method Describes recent strategies for fixed-length, dense word embeddings.
result Word embeddings encode syntactic and semantic information and improve NLP tasks.
Fixed angles of convex polygons lead to combinatorially rich polytopes.
problem Understanding the structure of convex polygons with fixed vertex angles.
method Combining combinatorial and geometric approaches, including dual polytopes and Schwarz-Christoffel maps.
result Fixed-angles polytopes are dual to cyclic polytopes under certain conditions.
Extends deep learning for hierarchical data to improve classification accuracy.
problem Classification with costly features in hierarchical data.
method Extended deep reinforcement learning with hierarchical deep sets and softmax.
result Superior performance on seven datasets, including malicious web domain classification.
The quality of machine translation is rapidly evolving. Today one can find several machine translation systems on the web that provide reasonable translations, although the systems are not perfect. In some specific domains, the quality may decrease. A recently proposed approach to this domain is neural machine translat…
OPORP combines permutation and random projection for efficient data vector compression.
problem Efficiently estimating cosine similarity in embedding-based retrieval applications.
method OPORP uses a permutation followed by a random vector dot product, then aggregates and normalizes the results into bins.
result OPORP improves the estimation of cosine similarity, reducing variance and improving accuracy.
Anomaly detection with LSTM neural networks for variable-length data.
problem Anomaly detection in variable-length data sequences.
method LSTM neural networks, OC-SVM, SVDD, gradient and quadratic programming.
result Significant performance gains over conventional methods.
TREP learns pedestrian trajectories efficiently without needing full datasets.
problem Learning fixed-length vector representations of variable-length trajectories.
method Actor-critic sequence-to-sequence autoencoder with spatial-aware objective function.
result TREP efficiently learns trajectory representations without needing full datasets.
Predicts path failures in evolving networks using deep learning.
problem Predicting path failures in time-evolving graphs.
method LRGCN, SAPE
result LRGCN outperforms other methods in path failure prediction.
Novel graph neural network combines random walks with local message passing.
problem Graph neural networks struggle with long-range dependencies.
method Combines random walks with local message passing in a novel architecture.
result Significant performance improvements on graph benchmarks.
Proposes SSM to improve CTR prediction with deep neural networks.
problem Improving CTR prediction with deep neural networks.
method Designs an orthogonal base convolution and pooling model to learn multi-scale base semantic representation.
result Demonstrates superior performance in CTR prediction.
Adaptive TFTs improve cryptocurrency price prediction accuracy.
problem Precise short-term price prediction in volatile cryptocurrency markets.
method Dynamic subseries lengths and pattern-based categorization.
result Significantly outperforms baseline models in prediction accuracy and profitability.
Characterizes curves for minimal surfaces in de Sitter space.
problem Minimal surfaces in de Sitter space.
method Variational problem to find critical points of center of mass.
result Curves are critical points of center of mass.
Study on dynamic curves with elastic energy and spontaneous curvature.
problem Modeling and analyzing dynamic planar curves with elastic energy.
method Gradient flow of inclination angle, nonlocal quasilinear system, local well-posedness, global existence, convergence.
result Local well-posedness, global existence, convergence of the flow for weak regularity initial data.
Moduli spaces of hyperbolic surfaces with geodesic boundary components of fixed lengths may be endowed with a symplectic structure via the Weil-Petersson form. We show that, as the boundary lengths are sent to infinity, the Weil-Petersson form converges to a piecewise linear form first defined by Kontsevich. The proof …
Motivated by the study of billiards in polygons, we prove fine results for the distribution of gaps of directions of saddle connections on translation surfaces. As an application we prove that for almost every holomorphic differential ω on a Riemann surface of genus g≥2 the smallest gap between saddle connecti…
End-to-end learning framework for tree-structured data.
problem Learning models struggle with tree-structured data due to lack of fixed-length vectors.
method Developed a novel framework for generic semantic tree-structured data of arbitrary topology.
result Framework yields comparable performance to standard models with dedicated feature-vectors and outperforms in compositional data.
Self multi-head attention improves speaker recognition for long utterances.
problem Speaker recognition for long speech segments using Deep Learning.
method Convolutional Neural Network (CNN) for short-term features, self multi-head attention for long-term embeddings.
result Self multi-head attention outperforms other pooling methods by 18% relative EER on VoxCeleb1 dataset.
Optimizes curves on Riemannian manifolds to minimize curvature.
problem Minimizing curvature on curves with fixed length and endpoints on Riemannian manifolds.
method Solves a second order ODE system derived from the optimization problem.
result Solutions to the optimization problem satisfy a second order ODE system.
If (M,g) is a compact Riemannian surface then the integrals of L2(M)-normalized eigenfunctions ej over geodesic segments of fixed length are uniformly bounded. Also, if (M,g) has negative curvature and γ(t) is a geodesic parameterized by arc length, the measures ej(γ(t))dt on R tend to zero in the …
Deep learning faces adoption challenges in business analytics.
problem Adoption of deep learning in business analytics is hindered by various factors.
method Empirical study based on three industry use cases.
result Gradient boosting is recommended for structured datasets in business analytics.
This paper classifies all planar p-elasticae and their properties.
problem Understanding the properties of p-elasticae in the plane.
method Introducing new types of p-elliptic functions to classify all planar p-elasticae.
result Explicit formulae and optimal regularity of all planar p-elasticae are obtained.
Study curves evolving by gradient flow of elastic energy, proving existence, smoothing, and convergence.
problem Evolution of curves with fixed length and clamped boundary conditions.
method Negative L2-gradient flow of elastic energy, existence, parabolic smoothing, constrained Lojasiewicz-Simon gradient inequality. result Convergence to a critical point as time tends to infinity.
End-to-end model extracts nested terms without extra features.
problem Automatic term extraction for nested terms.
method Deep learning model that predicts conceptual terms within fixed sentence lengths.
result High recall and comparable precision on term extraction task.