BOIL updates model body only, showing better few-shot learning performance.
problem Few-shot learning efficiency with model reuse vs. change.
method Proposes BOIL, updating only model body, freezing head.
result Significantly outperforms MAML on cross-domain tasks.
New findings on representation changes in transfer learning.
problem Understanding representation changes in transfer learning.
method Analysis of pre-trained and fine-tuned representations in transfer learning.
result Representation changes mainly occur in the head, but early layers are also important for cross-domain adaptation.
An important part of many machine learning workflows on graphs is vertex representation learning, i.e., learning a low-dimensional vector representation for each vertex in the graph. Recently, several powerful techniques for unsupervised representation learning have been demonstrated to give the state-of-the-art perfor…
New method learns useful disentangled representations from weakly labeled data.
problem Learning useful representations from weakly labeled data.
method Model pairs of non-i.i.d. images, learn disentangled representations without requiring annotation.
result Learn disentangled representations reliably from pairs of images without requiring group, individual factor, or number of changed factors annotation.
Paper proposes a method to improve circular coordinate representation for detecting changes in high-dimensional datasets.
problem Detecting changes in high-dimensional datasets with preserved topological structures.
method Adapt circular coordinate framework using a generalized penalty function instead of an L2 penalty.
result Circular coordinates with generalized penalty can detect changes in high-dimensional datasets under different sampling schemes.
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.
We train multi-task autoencoders on linguistic tasks and analyze the learned hidden sentence representations. The representations change significantly when translation and part-of-speech decoders are added. The more decoders a model employs, the better it clusters sentences according to their syntactic similarity, as t…
The paper develops a new formula for financial pricing under multiple interest rates and collateralization.
problem Financial pricing under multiple interest rates and collateralization.
method Derives a change of measure formula for recursive conditional expectations in a jump-diffusion setting.
result Generalizes the change of numéraire technique for multiple interest rates and collateralization.
Study reveals differences in medical image models' hidden representation refinement.
problem Understanding how intrinsic dimensionality changes in neural network hidden representations across different domains.
method Analysis of 11 natural and medical image datasets using 6 network architectures.
result Medical image models refine hidden representations earlier, suggesting differences in feature abstraction.
The paper analyzes MAML's representation using RSA, revealing that feature reuse is not the primary reason for its success.
problem Understanding why model-agnostic meta-learning (MAML) works well in few-shot learning tasks.
method Representation similarity analysis (RSA) applied to MAML's few-shot learning instantiation.
result Feature reuse is not the primary reason for MAML's success; instead, it is the learning task itself that increases representation similarity.
New framework to test neural network representation similarity measures.
problem Disagreements among dissimilarity measures in neural networks.
method Statistical testing framework to evaluate measures based on functional behavior.
result Current metrics have different weaknesses; a classical baseline performs surprisingly well.
New method combines domain changes and sparse mixing for better latent variable learning.
problem Challenges in identifying latent variables due to insufficient domain changes and violated sparsity constraints.
method Combines sufficient changes and sparse mixing constraints, using domain encoding networks and variational autoencoders.
result Identifiability of latent variables achieved with less restrictive constraints.
A new geometric metric identifies true data changes from parametrization artifacts in high-dimensional representations.
problem Quantifying representation drift in high-dimensional data using Euclidean or cosine distances can misattribute changes due to arbitrary parametrizations.
method Introducing the Fubini Study metric to identify representations that differ only by gauge transformations.
result The Fubini Study metric isolates intrinsic evolution by remaining invariant under gauge-induced fluctuations, providing a diagnostic for meaningful structural changes.
Autoencoder detects subtle changes in time series data.
problem Detect abrupt changes in time series data with high accuracy.
method Autoencoder with time-invariant representation and postprocessing.
result Outperforms baseline methods on various data sets.
This paper tackles causal representation learning from multiple distributions without hard interventions.
problem Recovering latent causal variables and their relations from multiple distributions.
method Develops general solutions for causal representation learning without hard interventions, under sparsity constraints and suitable change conditions.
result Recovering the moralized graph of the underlying directed acyclic graph and latent variables related to the underlying causal model.
The paper detects changes in graph signal means offline.
problem Segmenting and detecting changes in multivariate signals over graph nodes.
method Model selection approach exploiting sparsity in spectral domain.
result Proof of non-asymptotic oracle inequality for change-point detection.
We consider the problem of building a state representation model in a continual fashion. As the environment changes, the aim is to efficiently compress the sensory state's information without losing past knowledge. The learned features are then fed to a Reinforcement Learning algorithm to learn a policy. We propose to …
Facial attribute editing aims to manipulate single or multiple attributes of a face image, i.e., to generate a new face with desired attributes while preserving other details. Recently, generative adversarial net (GAN) and encoder-decoder architecture are usually incorporated to handle this task with promising results.…
A crucial challenge in image-based modeling of biomedical data is to identify trends and features that separate normality and pathology. In many cases, the morphology of the imaged object exhibits continuous change as it deviates from normality, and thus a generative model can be trained to model this morphological con…
New method recovers causal DAGs from general environments without strict assumptions.
problem Recovering causal DAGs from real-world data with varying distributions.
method Formalizes desiderata for causal representation learning in general environments, leveraging sufficient change conditions up to third-order derivatives.
result Fully recovers latent DAG and identifies latent variables up to minor indeterminacies under nonparametric mixing.
When a boudnary-parabolic representation of a link group to PSL(2,C) is given, Inoue and Kabaya suggested a combinatorial method to obtain the developing map of the representation using the octahedral triangulation and the shadow-coloring of certain quandle. Quandle is an algebraic system closely related wit…
AdaRL adapts quickly to new environments with minimal data.
problem Quickly adapting to new environments in reinforcement learning.
method AdaRL uses a parsimonious graphical representation to encode changes across domains.
result AdaRL can efficiently adapt policies to target domains with few samples.
Innovative inequalities for divergences with applications in PAC-Bayesian bounds and Monte Carlo.
problem Developing new inequalities for divergences.
method Introducing novel change of measure inequalities for f-divergences and α-divergences. result Applications in PAC-Bayesian bounds and Monte Carlo estimates.
Tree-based regularization improves latent variable inference from related datasets.
problem Inferring latent variables from multiple related datasets in causal systems.
method Tree-Based Regularization (TBR) for sparse changes across environments.
result TBR identifies true latent variables up to simple transformations under sparse changes.
New framework TDRL identifies latent causal variables from sequential data.
problem Identify latent causal variables from sequential data.
method Proposes TDRL framework to recover time-delayed latent causal variables and identify their relations from measured sequential data.
result Identifies latent causal variables reliably from sequential data.
Model predicts travel time under rare conditions using a vector-space model.
problem Predicting travel time under rare temporal conditions (e.g., holidays, school vacations) is challenging due to limited historical data and other temporal changes.
method Presented a vector-space model for encoding rare temporal conditions, allowing coherent representation learning across different conditions.
result Increased performance for travel time prediction over different baselines when using the vector-space encoding for representing the temporal setting.
Improves contrastive learning invariance with novel training objectives and feature averaging.
problem Contrastive learning's implicit invariance is insufficient for robust performance.
method Introduces a novel training objective and feature averaging approach to enforce invariance.
result Improved performance and robustness to transformations on downstream tasks.
Study on bending knots and energy changes in 3D space.
problem Understanding energy changes in knots under small deformations.
method Analyzes infinitesimal bending of knots and energy changes using Willmore and Mobius energies.
result Changes in energy under small deformations of knots have been quantified.
Unified framework for disentangled representations using mechanistic independence.
problem Identifiability of disentangled latent factors under statistical dependencies.
method Introduces mechanistic independence to characterize latent factors by their actions on observed variables, proposing various independence criteria.
result Establishes conditions for identifiability of latent subspaces without statistical assumptions.
Paper proposes semi-supervised learning using change points for sequence classification.
problem Limited labeled data for sequential sensor data classification.
method Change point detection for identifying class changes, semi-supervised learning with labeled and unlabeled data.
result Improved classification performance on human activity recognition datasets.
Simplified trust region method reduces representation change during fine-tuning.
problem Stability and representational collapse in fine-tuning pre-trained models.
method Replaces adversarial objectives with parametric noise in trust region theory.
result Matches or exceeds previous trust region methods in performance and speed.
New method extracts brain age from MRI sequences over time.
problem Lack of ground-truth labels in longitudinal neuroimaging data.
method Combines factor disentanglement with self-supervised learning.
result Extracts brain age information from MRI sequences.
In the artificial intelligence field, learning often corresponds to changing the parameters of a parameterized function. A learning rule is an algorithm or mathematical expression that specifies precisely how the parameters should be changed. When creating an artificial intelligence system, we must make two decisions: …
Nowadays, with the availability of massive amount of trade data collected, the dynamics of the financial markets pose both a challenge and an opportunity for high frequency traders. In order to take advantage of the rapid, subtle movement of assets in High Frequency Trading (HFT), an automatic algorithm to analyze and …
A connection is made between the Krammer representation and the Birman-Murakami-Wenzl algebra. Inspired by a dimension argument, a basis is found for a certain irrep of the algebra, and relations which generate the matrices are found. Following a rescaling and change of parameters, the matrices are found to be identica…
Better data representations can simplify learning tasks by aligning model distributions with true data distributions.
problem Learning complexity influenced by the alignment of model distributions with true data distributions.
method Analyzed the effect of data representations on learning complexity using a task complexity score and information coding length.
result Better representations can simplify learning tasks by aligning model distributions with true data distributions, improving learning outcomes.
Paper presents a new way to estimate model changes without full model evaluation.
problem Efficiently estimating changes in model parameters and outputs due to data point removal.
method Dual representation of influence functions for linearizable models, reducing computational complexity.
result The dual representation can be an efficient alternative to original influence functions, especially for large models.
We consider the problem of building a state representation model for control, in a continual learning setting. As the environment changes, the aim is to efficiently compress the sensory state's information without losing past knowledge, and then use Reinforcement Learning on the resulting features for efficient policy …
Recent progress in recommender system research has shown the importance of including temporal representations to improve interpretability and performance. Here, we incorporate temporal representations in continuous time via recurrent point process for a dynamical model of reviews. Our goal is to characterize how change…
Integrates inductive biases into VAEs using intermediary latent variables.
problem Ineffective mechanisms for incorporating inductive biases into VAEs.
method InteL-VAEs use an intermediary latent space to control encoding, with a parametric function to enforce desired properties.
result InteL-VAEs lead to better generative models and representations.
Stability is a key aspect of data analysis. In many applications, the natural notion of stability is geometric, as illustrated for example in computer vision. Scattering transforms construct deep convolutional representations which are certified stable to input deformations. This stability to deformations can be interp…
Humans and animals show remarkable flexibility in adjusting their behaviour when their goals, or rewards in the environment change. While such flexibility is a hallmark of intelligent behaviour, these multi-task scenarios remain an important challenge for machine learning algorithms and neurobiological models alike. We…
We survey generalisations of the Chang-Skjelbred Lemma for integral coefficients. Moreover, we construct examples of manifolds with actions of tori of rank > 2 whose equivariant cohomology is torsion-free, but not free. This answers a question of Allday's. The "mutants" we construct are obtained from compactified repre…
We present a new and easy-to-implement sequential sampling method for CGMY processes with either finite or infinite variation, exploiting the time change representation of the CGMY model and a decomposition of its time change. We find that the time change can be decomposed into two independent components. While the fir…
This paper addresses the problem of change-point detection on sequences of high-dimensional and heterogeneous observations, which also possess a periodic temporal structure. Due to the dimensionality problem, when the time between change-points is on the order of the dimension of the model parameters, drifts in the und…
New lattice path method for statistical inference of persistent diagrams.
problem Statistical inference on persistent diagrams.
method Lattice path representation and combinatorial enumerations.
result Topological changes observed in spike proteins of COVID-19 virus.
Proposes FSM-IRL to learn invariant network representations considering feature and structural shifts.
problem Spatial heterogeneity and temporal dynamics lead to OOD generalization issues in geographic networks.
method Introduces FSM-IRL model that accounts for feature and structural distribution shifts using causal attention and reweighting.
result Demonstrates strong learning capabilities on geographic and social network datasets in OOD scenarios.
Machine learning for healthcare often trains models on de-identified datasets with randomly-shifted calendar dates, ignoring the fact that data were generated under hospital operation practices that change over time. These changing practices induce definitive changes in observed data which confound evaluations which do…