Adaptive computation improves machine reasoning for complex tasks.
problem Learning to perform multi-hop inference for complex tasks.
method Introduced a model with Adaptive Computation Time to learn the number of inference steps.
result Adaptive computation provides a small performance benefit and insight into reasoning.
New method uses fewer parameters to match state-of-the-art performance on multiple natural language tasks.
problem Efficiently adapting BERT for multiple tasks with fewer parameters.
method PALs (projected attention layers) for shared BERT model with task-specific parameters.
result Matches state-of-the-art performance on GLUE benchmark with 7 times fewer parameters.
End-to-end ASR error detection using audio-transcript entailment.
problem Detecting transcription errors in ASR systems to prevent error propagation.
method Proposes a novel end-to-end approach using audio-transcript entailment, with acoustic and linguistic encoders.
result Achieves CER of 26.2% on all transcription errors and 23% on medical errors specifically, improving by 12% and 15.4% respectively over a strong baseline.
Optimizes diversification in catastrophe risk pooling using asymptotic analysis.
problem Maximizing diversification benefit from catastrophic events in insurance pools.
method Asymptotic analysis to solve high-dimensional optimization problem.
result Derives an asymptotically optimal pool that approximates practical optimal pool.
The paper proposes methods to control errors in language generation models using textual entailment.
problem The lack of a correctness metric hinders applying principled methods to language generation tasks.
method The paper leverages textual entailment to evaluate correctness and proposes two selective generation algorithms: SGen^Sup and SGen^Semi.
result The proposed algorithms control the false discovery rate with respect to textual entailment and achieve comparable selection efficiency to baselines.
New function class characterizes loss landscape of deep neural networks without over-parametrization.
problem Complex loss landscape of deep neural networks without over-parametrization.
method Proposed a novel class of functions to characterize loss landscape without over-parametrization.
result Gradient-based optimizers possess theoretical guarantees of convergence under the new function class assumption.
Approaches KL divergence for learning multi-sense word distributions.
problem Capturing the polysemy and uncertainty of words in word embeddings.
method Modeling words as multi-sense Gaussian mixtures and using KL divergence for learning.
result The proposed approach effectively captures word entailment and distribution similarity.
New word distributions capture multiple meanings and outperform existing methods.
problem Capturing semantic information for words with multiple meanings.
method Gaussian mixtures with an energy-based max-margin objective.
result Multimodal word distributions outperform word2vec and Gaussian embeddings.
Algorithm for recognizing and performing Reidemeister moves in Gauss diagrams.
problem Recognizing and performing Reidemeister moves in Gauss diagrams.
method Simple algorithm for recognizing and performing Reidemeister moves in Gauss diagrams.
result Simple algorithm for recognizing and performing Reidemeister moves in Gauss diagrams.
Trivial links are unique up to number of link components, but they can be hard to recognize from arbitrary diagrams. We define a new measure of the complexity of a link embedding, the crumple, and show how this may be used to measure progress toward a trivial embedding. In conjunction with a modified form of arc presen…
Horovod simplifies multi-GPU training in TensorFlow.
problem Efficient multi-GPU training in TensorFlow with minimal code changes.
method Efficient inter-GPU communication via ring reduction and minimal code modifications.
result Faster, easier distributed training in TensorFlow.
The paper analyzes sports commentary to automatically recognize events and extract insights.
problem Automatically recognizing and categorizing major actions in sports events from commentary.
method Used multiple Natural Language Processing techniques for classification and sentiment analysis.
result Identified insights from analyzing live sport commentaries and classifying major actions.
Analyzes surfaces with bounded curvature, proving properties and continuity.
problem Analyzes surfaces with locally bounded integral curvature.
method Analyzes surfaces as metric measure spaces, proving infinitesimal Hilbertianity, local doubling, and Poincaré inequality.
result Proves existence of jointly Hölder continuous heat kernel for Cheeger Laplacian.
Analyzes surfaces with bounded curvature, proving properties and existence of heat kernels.
problem Analyzes surfaces with locally bounded integral curvature.
method Analyzes surfaces as metric measure spaces, proving infinitesimal Hilbertianity, local doubling, and Poincaré inequality.
result Existence of jointly Hölder continuous heat kernel for Cheeger Laplacian.
Challenge aims to recognize music genres from audio.
problem Recognizing music genres from audio recordings.
method Open data challenge with submissions evaluated.
result Results presented from the challenge.
This is an expanded and updated version of a lecture series I gave at Seoul National University in September 1997. It is in some sense an update of the 1979 Griffiths and Harris paper with a similar title. I discuss: Homogeneous varieties, Topology and consequences Projective differential invariants, Varieties with deg…
A conceptor-based approach helps robots recognize human internal states.
problem Recognizing human internal states for diagnostic purposes in autism therapy.
method A conceptor-based classifier to classify internal states.
result Initial results show potential for detailed diagnostic information.
We embed directed acyclic graphs using hyperbolic spaces and geodesic cones.
problem Learning graph representations that preserve hierarchical structure.
method Use hyperbolic spaces and geodesic cones to define embeddings of directed acyclic graphs.
result Our method significantly outperforms existing approaches in graph representation learning.
Hierarchical density embeddings capture word relationships with uncertainty.
problem Capturing semantic relationships and uncertainty in word embeddings.
method Learn hierarchical representations through probability density encapsulation, using simple loss functions and distance metrics.
result State-of-the-art performance on WordNet and Hyperlex datasets.
The Trek Separation Theorem (Sullivant et al. 2010) states necessary and sufficient conditions for a linear directed acyclic graphical model to entail for all possible values of its linear coefficients that the rank of various sub-matrices of the covariance matrix is less than or equal to n, for any given n. In this pa…
Study uses CNN and LSTM to recognize stock chart patterns.
problem Recognizing stock chart patterns for trading.
method Used CNN and LSTM neural networks on historical stock data.
result Obtained accuracies for recognizing two common chart patterns.
Deep learning models accurately recognize and estimate physical activity types and energy expenditure from wrist accelerometer data.
problem Rigorous evaluation of wrist-worn accelerometers for assessing physical activity across the lifespan.
method Built deep learning networks to extract spatial and temporal representations from time-series data, recognizing physical activity types and estimating energy expenditure.
result Deep learning models achieved high performance: F1 scores of 0.82, 0.81, and 95 for sedentary, locomotor, and lifestyle activities, respectively; root mean square error of 1.1 for EE estimation.
Adyan and Rabin showed that most properties of groups cannot be algorithmically recognized from a finite presentation alone. We prove that, if one is also given a solution to the word problem, then the class of fundamental groups of closed, geometric 3-manifolds is algorithmically recognizable. In our terminology, the …
The paper categorizes music emotions and improves music retrieval.
problem Inefficient music retrieval based on album information.
method Categorical emotion expression, Fisher's separation theorem, feature extraction, Support Vector Machines.
result Maximum separability occurs between relaxing and epic music parts.
To a closed braid in a solid torus we associate a trace graph in a thickened torus in such a way that closed braids are isotopic if and only if their trace graphs can be related by trihedral and tetraherdal moves. For closed braids with a fixed number of strands, we recognize trace graphs up to isotopy and trihedral mo…
TOQ-Nets learn to recognize complex temporal events with varying objects and sequences.
problem Recognizing complex relational-temporal events with varying numbers of objects and sequence lengths.
method Neuro-symbolic networks with reasoning layers for finite-domain quantification over objects and time.
result TOQ-Nets can generalize to scenarios with more objects than training data and temporal warpings.
WEEND uses a neural network to recognize speech and assign speakers to words.
problem End-to-end neural diarization without additional ASR and orchestration.
method Multi-task learning with an auxiliary network for ASR and speaker diarization.
result WEEND outperforms turn-based diarization and can handle 5-minute audio.
Extends cutting and blowing up to nonrational symplectic settings.
problem Nonrational symplectic toric structures.
method Cutting and blowing up in nonrational directions.
result Extension to symplectic toric manifolds and orbifolds.
Paper tackles incremental few-shot learning with novel classes.
problem Learning new classes with limited data and without re-training.
method Attention Attractor Network (AAN) for incremental few-shot learning.
result AAN helps recognize new classes without forgetting old ones.
HHAR-net uses neural networks to recognize human activities at different levels of abstraction.
problem Recognizing different layers of human activities concealed in behavior.
method Hierarchical classification with Neural Networks.
result 95.8% accuracy for low-level activities and 92.8% overall accuracy.
This paper compares machine learning methods for recognizing lane change intentions from vehicle trajectories.
problem Accurately detecting and predicting lane change processes in autonomous vehicles.
method Comparison of different machine learning methods on high-dimensional time series data.
result Ensemble methods reduce Type II and Type III classification errors, while LightGBM outperforms XGBoost in training efficiency.
RaRecognize learns to recognize rare classes in a stream of data.
problem Learning to recognize rare classes in a continuous stream of data.
method Estimates a general decision boundary, learns individual rare subclasses, flags new subclasses.
result RaRecognize outperforms state-of-the-art baselines on real-world datasets.
Researchers develop hyperbolic neural networks for improved data embedding.
problem Lack of hyperbolic neural network layers limits the use of hyperbolic embeddings in machine learning.
method Combining Möbius gyrovector spaces and Riemannian geometry of the Poincaré model to derive hyperbolic neural network layers.
result Hyperbolic sentence embeddings outperform or match Euclidean variants on textual entailment and noisy-prefix recognition tasks.
R package for multi-objective model selection in statistics.
problem Model selection challenges in statistics, especially for penalized models.
method Multi-objective optimization using Gaussian process-based optimization.
result Identification of hyperparameter values that represent desirable trade-offs.
This paper is an attempt at understanding the quantum-like dynamics of financial markets in terms of non-differentiable price-time continuum having fractal properties. The main steps of this development are the statistical scaling, the non-differentiability hypothesis, and the equations of motion entailed by this hypot…
We prove that a wide class of correlated stochastic volatility models exactly measure an empirical fact in which past returns are anticorrelated with future volatilities: the so-called ``leverage effect''. This quantitative measure allows us to fully estimate all parameters involved and it will entail a deeper study on…
This paper improves facial expression recognition using CNNs and coherence constraints.
problem Facial expression recognition from static images and video sequences is challenging.
method Investigates the use of Convolutional Neural Networks (CNNs) with coherence constraints in a semi-supervised setting.
result Coherence constraints improve facial expression recognition quality, especially in the presence of occlusions.
Paper develops deep learning for signal recognition in long perimeter fiber optic sensors.
problem Difficult signal-jamming environments and stringent error requirements.
method Two-level event detection architecture with ensemble of deep convolutional networks.
result Efficient and robust multiclass detection algorithms with high adaptability.
COBRA efficiently clusters data using pairwise constraints with minimal queries.
problem Clustering datasets with user-defined constraints.
method Over-clusters data with K-means, then merges clusters based on constraints.
result COBRA outperforms state-of-the-art methods in clustering quality and runtime.
Paper presents an audiovisual model to recognize sounds from weakly labeled video data.
problem Sound recognition from weakly labeled video data.
method Audiovisual fusion model with attention mechanism.
result The model achieves a mean Average Precision (mAP) of 46.16 on AudioSet, outperforming state-of-the-art models.
Complexity class determined for recognizing torus knots.
problem Recognizing specific torus knots and related knots.
method Based on recent work on detecting knottedness.
result Recognition problem is in NP and co-NP.
System filters and ranks medical answers using pre-trained models.
problem Challenges in ranking and classifying medical answers due to input size and dataset limitations.
method Multi-task learning with pre-trained models as feature extractors.
result Achieved high performance on medical QA task (Spearman's Rho 0.338, MRR 0.9622).
We study online prediction of bounded stationary ergodic processes. To do so, we consider the setting of prediction of individual sequences and build a deterministic regression tree that performs asymptotically as well as the best L-Lipschitz constant predictors. Then, we show why the obtained regret bound entails the …
Recognizing group activities is challenging due to the difficulties in isolating individual entities, finding the respective roles played by the individuals and representing the complex interactions among the participants. Individual actions and group activities in videos can be represented in a common framework as the…
We present a simple-to-apply criterion for recognizing topological groups that are (locally) homeomorphic to LF-spaces.
The Killing operator on a Riemannian manifold is a linear differential operator on vector fields whose kernel provides the infinitesimal Riemannian symmetries. The Killing operator is best understood in terms of its prolongation, which entails some simple tensor identities. These simple identities can be viewed as aris…
SpatialSim benchmarks machine learning in recognizing object spatial configurations.
problem Machine learning in recognizing precise geometrical configurations of groups of objects.
method SpatialSim benchmark with tasks of Identification and Comparison, using Graph Neural Networks (MPGNNs).
result MPGNNs outperform baselines in recognizing spatial configurations, highlighting current limits.
A new framework for sparse and structured neural attention.
problem Improving interpretability and performance of neural networks.
method Proposes a smoothed max operator framework for attention mechanisms.
result Improved interpretability without sacrificing performance.