DeepCCG adapts classifiers to representation shifts in one step.
problem Adapting classifiers to shifts in continuous representation.
method Empirical Bayesian approach using class conditional Gaussian classifier and KL divergence for selection.
result DeepCCG reduces performance change due to representation shifts.
Classifies SL(2;C) representations of a Brieskorn homology 3-sphere.
problem Classifying SL(2;C) representations of a Brieskorn homology 3-sphere.
method Shows irreducible representations are conjugate to SU(2) or SL(2;R) representations. Constructs SL(2;R) representations from PSL(2;R) representations of the orbifold.
result Classifies SL(2;C) representations and provides a construction method.
Enhances multi-tag classification using low-dimensional vector representations and virtual data.
problem Improving the performance of multi-tag classifiers.
method Embedding raw data into a low-dimensional feature space, then generating virtual data from linear operations on these vectors, to train multi-tag classifiers.
result Significant improvement in F1 scores (up to 224%) compared to training directly with raw data.
Classifies Lie group representations linked to quaternion-Kähler symmetric spaces.
problem Classifying representations of Lie groups with specific orbit spaces.
method Analyzing representations of Sp(1)k-extensions and quaternion-Kähler symmetric spaces. result Representations are derived from isotropy representations of quaternion-Kähler symmetric spaces.
ConvNet classifies whale vocalizations and ambient noise in acoustic recordings.
problem Automated detection and classification of marine mammal vocalizations in acoustic recordings.
method Convolutional Neural Network with a novel acoustic representation.
result Classifier accurately detects and classifies whale vocalizations and ambient noise.
Classifies extensions of gauge Lie algebras for positive energy representations.
problem Classifying continuous central extensions of gauge Lie algebras for positive energy representations.
method 1-parameter group of transformations of the base manifold.
result Classification of extensions that are covariant under transformations.
PLIs improve classifier performance by fine-tuning latent representations.
problem Difficult interpretation of high-dimensional latent representations in neural networks.
method Back-propagation of manual changes to low-dimensional embeddings using t-distributed stochastic neighbourhood embeddings.
result Manual separation of class clusters in latent space enhances classifier performance.
Enhances generative models with classifier representations.
problem Improving the quality of generative models.
method Discriminative regularization applied to variational autoencoders.
result Generated samples are clearer and have higher visual quality.
Study of vector bundles over classifying spaces for infinite discrete groups.
problem Understanding vector bundles over classifying spaces for infinite discrete groups.
method Homotopy theoretical framework for infinite discrete groups, relating to Novikov conjecture.
result Established a connection between representation spaces and vector bundles over classifying spaces.
Develops NPSVC++ to improve NPSVC performance through representation learning.
problem Feature suboptimality and class dependency in NPSVC training.
method Multi-objective optimization and end-to-end learning of NPSVC and its features.
result NPSVC++ ensures feature optimality across classes, overcoming training issues.
Proposes Decodable Information Bottleneck for optimal representation learning.
problem Finding optimal representations for supervised learning.
method Integrates information retention and compression with the desired predictive family.
result Optimal representations lead to better expected test performance and can be estimated with guarantees.
Classifies Zariski closures of positive representations in Lie groups.
problem Classifying Zariski closures of positive representations in Lie groups.
method Classifies the Lie algebra of the Zariski closure of a discrete subgroup with specific properties.
result Obtains a new proof of Guichard's classification of Zariski closures of Hitchin representations.
This belongs to a series of papers devoted to the study of the cohomology of classifying spaces of Lie groupoids. Our aim here is to introduce and study the notion of representation up to homotopy of Lie groupoids, the resulting derived category, and to show that the adjoint representation is well defined as a represen…
Classifies representations up to dimension 3g-3 for surface mapping class groups.
problem Classifying representations of mapping class groups up to a certain dimension.
method Direct sum of a 2g or 2g+1 dimensional representation and a trivial one.
result Any representation up to dimension 3g-3 is a direct sum of a 2g or 2g+1 dimensional representation and a trivial one.
New bound proves rationality helps generalization in self-supervised learning.
problem Proving generalization gap in self-supervised learning.
method Proving upper bound on generalization gap for classifiers using rationality and self-supervised representations.
result Generalization gap tends to zero if classifier complexity is low relative to number of samples.
This paper classifies equivariant principal bundles over a 2-sphere using isotropy representations.
problem Classifying equivariant principal bundles over the 2-sphere.
method Using isotropy representations to classify bundles over the 2-sphere.
result Equivariant principal bundles over the 2-sphere can be classified by a Γ-fixed set of homotopy classes of maps and first Chern class.
Study finds open data sets favor Western locales, impacting classifier performance.
problem Impact of biased open data sets on classifier performance in the developing world.
method Analysis of two large, publicly available image data sets and classifiers trained on them.
result Open data sets exhibit a bias towards Western locales, affecting classifier performance.
Paper proposes CCVR to improve classifier calibration in federated learning with non-IID data.
problem Improving classifier calibration in federated learning with non-IID data.
method Proposes CCVR, a simple algorithm that adjusts the classifier using virtual representations.
result CCVR achieves state-of-the-art performance on popular federated learning benchmarks.
A simple method flags images as out-of-distribution based on their distance to nearest neighbors.
problem Detecting images not aligned with a trained model's in-distribution data.
method Flag images as OOD if their average distance to K nearest neighbors is large in the classifier's representation space.
result Simple methods can outperform more complex ones when considering learned representations.
For random graphs distributed according to stochastic blockmodels, a special case of latent position graphs, adjacency spectral embedding followed by appropriate vertex classification is asymptotically Bayes optimal; but this approach requires knowledge of and critically depends on the model dimension. In this paper, w…
This paper classifies tweets into positive and negative sentiments using distributed word and sentence representations.
problem Classifying tweets into positive and negative sentiments.
method Used distributed representations of words and sentences, and LSTM and CNN networks for classification.
result Achieved accuracies as high as 81%.
The paper extends calibration to sets of probabilistic classifiers, finding many ensembles are poorly calibrated.
problem Evaluating the validity of epistemic uncertainty in sets of probabilistic classifiers.
method Proposed a novel nonparametric calibration test for sets of probabilistic classifiers.
result Ensembles of deep neural networks are often not well calibrated.
Classifies finite orbits of mapping class group actions on surface group representations.
problem Finite orbits of mapping class group actions on surface group representations.
method Classification based on surface genus and representation properties.
result Finite orbits correspond to homomorphisms with finite image for genus at least two, and to finite or special dihedral representations for genus one.
Paper tackles deconfounding age effects in dementia detection models.
problem Dementia detection models are affected by age, leading to potential non-generalizable accuracies.
method Proposes fair representation learning to learn age-invariant representations.
result Best models compromise accuracy by only 2.56% and 1.54% on clinical datasets.
Generative classifier derived from any discriminative classifier rejects illegal inputs.
problem Detecting and rejecting illegal inputs like adversarial examples and out-of-distribution samples.
method SDIM-logit: learns generative classifier from logits of any discriminative classifier, imposing statistical constraints.
result SDIM-logit inherits performance of base classifier without loss and can reject illegal inputs.
LEEP measures transferability of learned representations efficiently.
problem Evaluating the transferability of learned representations in machine learning.
method LEEP: Log Expected Empirical Prediction, a simple measure requiring one pass through the target data set.
result LEEP predicts transfer and meta-transfer learning performance and convergence speed, outperforming existing measures.
Classifies prime algebraic tangles up to 14 crossings.
problem Classifying prime algebraic tangles systematically.
method Developed a novel canonical representation to distinguish mutant tangles.
result Increased classification of prime tangles up to 14 crossings.
Classifies invariant measures on specific character varieties.
problem Classifying invariant probability measures on character varieties.
method Measure disintegration along transverse Lagrangian tori fibrations.
result Ergodic measures are either counting measures on finite orbits or Liouville measures.
Defines spherical type surfaces via support function and classifies them.
problem Characterizing surfaces via support function.
method Weierstrass type representation for SS-surfaces with prescribed Gauss map.
result Every compact and connected SS-surface is the sphere.
iCaRL learns new classes incrementally from data streams.
problem Incremental learning of new concepts over time from a stream of data.
method iCaRL: learns strong classifiers and data representation simultaneously, allowing progressive addition of new classes.
result iCaRL can learn many classes incrementally over a long period of time, unlike other methods.
This paper investigates how data augmentation improves linear separation of manifold data.
problem Understanding how data augmentation enhances linear separation of manifold data.
method Investigates the conditions under which self-supervised representations can linearly separate multi-manifold data.
result Self-supervised learning can linearly separate manifolds with a smaller distance than unsupervised learning.
New algorithms improve time series classification accuracy and efficiency while enhancing interpretability.
problem Lack of interpretability in time series classification algorithms.
method Combining multiple resolutions and domains, using SEQL with greedy feature selection.
result SAX-SFA-SEQL achieves similar accuracy to state-of-the-art methods but with lower computational time.
TaCo prevents non-linear classifiers from detecting sensitive attributes.
problem Ensuring fairness in NLP models by preventing sensitive attribute detection.
method Targeted Concept Erasure (TaCo) removes sensitive information from final latent representations, even against non-linear classifiers.
result TaCo outperforms state-of-the-art methods in reducing sensitive attribute prediction accuracy while preserving overall task performance.
Develops a fair classifier for deep learning models.
problem Ensuring fairness in classification models across different sub-populations.
method Applies Rawlsian principles to minimize error rate on the worst-off sub-population.
result Introduces a practical method to adapt any black-box deep learning model to be fair.
Adversarial methods fail to remove demographic attributes from text data.
problem Demographic information of authors is encoded in text-based neural classifiers.
method Adversarial training to remove demographic attributes.
result Post-hoc classifiers trained on encoded sentences still achieve higher accuracies.
Neural-style algorithm classifies artistic styles of paintings.
problem Classifying artistic styles of paintings using neural networks.
method Uses neural-style algorithm to merge artistic style with content of images.
result Neural-style representation effectively classifies artistic styles of paintings.
In this paper we study weakly irreducible holonomy representations of the normal connection of a spacelike submanifold in a pseudo-Riemannian space from. We associate screen representations to weakly irreducible normal holonomy groups and classify the screen representations having the Borel-Lichnérowicz property. In pa…
In this paper we classify the reducible representations of compact simple Lie groups all of whose orbits are tautly embedded in Euclidean space with respect to Z_2 coefficients.
We classify irreducible representations of connected compact Lie groups whose orbit space is isometric to the orbit space of a representation of a finite extension of (positive dimensional) toric group. They turn out to be exactly the non-polar irreducible representations preserving an isoparametric submanifold and act…
We construct projective unitary representations of the smooth Deligne cohomology group of a compact oriented Riemannian manifold of dimension 4k+1, generalizing positive energy representations of the loop group of the circle. We also classify such representations under a certain condition. The number of the equivalence…
A neural network learns MRI scan-invariant features for brain tissue classification.
problem Lack of generalization in voxelwise classification methods due to scanner differences.
method Siamese neural network (MRAI-net) to learn acquisition-invariant representations.
result Linear classifier outperforms CNNs on limited training data for tissue classification.
Theory explains how linear networks can classify objects from neural responses.
problem Understanding how linear networks classify objects from neural responses.
method Characterized the perceptron's ability to classify objects from variable neural responses.
result The perceptron's capacity depends on the object manifold's dimensionality, size, and shape.
The paper classifies and proves properties of symmetry breaking operators for specific groups.
problem Classifying and understanding symmetry breaking operators for de Sitter and Lorentz groups.
method Constructing and classifying differential symmetry breaking operators, proving localness, and showing sporadic nature.
result All symmetry breaking operators are differential and sporadic, not obtainable by residue formulas.
Study classifies semi-discrete linear Weingarten surfaces with Weierstrass-type representations and analyzes their singularities.
problem Characterizing semi-discrete linear Weingarten surfaces with Weierstrass-type representations and their singularities.
method Established properties, classified, and analyzed the singularities of semi-discrete linear Weingarten surfaces in Riemannian and Lorentzian spaceforms.
result Defined and analyzed the singularities of semi-discrete linear Weingarten surfaces, including those with non-zero constant Gaussian curvature, parallel surfaces of minimal and maximal surfaces, and constant mean curvature 1 surfaces in de Sitter 3-space.
C2VAE learns disentangled and coupled representations without prior knowledge.
problem Learning disentangled and coupled representations in latent space.
method Introduces C2VAE, a self-supervised VAE that factorizes posterior and uses Gaussian copula for dependencies. result Demonstrates strong effect in enhancing disentangled representation learning.
Temporal networks are ubiquitous and evolve over time by the addition, deletion, and changing of links, nodes, and attributes. Although many relational datasets contain temporal information, the majority of existing techniques in relational learning focus on static snapshots and ignore the temporal dynamics. We propose…
We investigate orthogonal representations of compact Lie groups from the point of view of their quotient spaces, considered as metric spaces. We study metric spaces which are simultaneously quotients of different representations and investigate properties of the corresponding representations. We obtain some structural …
Study classifies Lie group representations with non-empty boundary orbit space.
problem Classifying representations of Lie groups with non-empty boundary orbit space.
method Detailed calculations based on previous work.
result Classification of Lie group representations with non-empty boundary orbit space.