Study finds Bitcoin crypto currency rate correlates with social network activity.
problem Detecting correlation between crypto currency rate and social network activity.
method Comparative correlation and fractal analysis of time series data.
result Time series of Bitcoin rate and social network activities exhibit self-similar and multifractal properties.
Study shows how correlations between neural activity affect classification capacity.
problem Understanding how correlations between neural activity impact classification performance.
method Calculated the capacity of neural activity on spherical manifolds with and without correlations between centroids and axes.
result Introducing correlations between neural activity centroids pushes spheres closer together, while correlations between axes shrink their radii, revealing a duality between correlations and geometry in classification.
This paper benchmarks Bayesian models' ability to estimate predictive correlations, especially for active learning.
problem Benchmarking how accurately Bayesian models estimate predictive correlations, especially in active learning.
method Considered transductive active learning as a benchmark, introduced meta-correlations and cross-normalized likelihoods.
result Meta-correlations and cross-normalized likelihoods can efficiently evaluate predictive correlations and are consistent with TAL performance.
Active learning optimizes correlation clustering by querying the most informative pairwise comparisons.
problem Efficiently clustering data with limited pairwise similarity information.
method Developed principled active learning approach using information-theoretic acquisition functions.
result Significantly outperforms existing baselines in clustering accuracy and query efficiency.
Paper develops active learning for clustering unknown pairwise similarities.
problem Learning positive and negative pairwise similarities efficiently.
method Generic active learning framework for correlation clustering.
result Demonstrates effectiveness of query strategies in clustering.
Motivated by social balance theory, we develop a theory of link classification in signed networks using the correlation clustering index as measure of label regularity. We derive learning bounds in terms of correlation clustering within three fundamental transductive learning settings: online, batch and active. Our mai…
We consider a few quantities that characterize trading on a stock market in a fixed time interval: logarithmic returns, volatility, trading activity (i.e., the number of transactions), and volume traded. We search for the power-law cross-correlations among these quantities aggregated over different time units from 1 mi…
Cryptocurrency market activity is decomposed into recurring and noise components, revealing patterns tied to macroeconomic reports.
problem Investigating temporal patterns of cryptocurrency market activity.
method Decomposition of market activity measures into recurring and noise components via correlation matrix formalism.
result Recurring market activity bursts coincide with significant U.S. macroeconomic reports, indicating their influence.
This work optimizes clustering with adaptive queries to minimize disagreements.
problem Minimizing disagreements in clustering with adaptive similarity queries.
method Active learning algorithms and information-theoretical bounds.
result Achieves an almost optimal trade-off between queries and clustering error.
DORA analyzes deep neural networks' internal representations to detect spurious correlations.
problem Detecting spurious correlations in deep neural networks' internal representations.
method DORA uses Extreme-Activation (EA) distance measure to assess representation similarities.
result Identifies internal representations capable of detecting spurious correlations.
Mean representations of VAEs are correlated but still useful for tasks.
problem Correlation between mean and sampled representations of VAEs.
method Selective posterior collapse to identify active and passive variables.
result Passive variables in mean representations are correlated but uncorrelated in sampled ones.
Gradient descent aligns neural feature matrices with pre-activation tangent features.
problem Understanding neural feature learning mechanisms.
method Analytical proof of alignment between weight matrices and pre-activation tangent features.
result Derivative alignment occurs almost surely in high-dimensional settings.
Wavelet analysis reveals financialization effects on oil-food price correlation.
problem Investigating the correlation between oil and food prices and their determinants.
method Wavelet analysis and energy-based measures to differentiate high and low frequency movements.
result Significant local correlation between food and oil is due to financialization and emerging economies' demand.
New findings connect shaped and unshaped neural networks using differential equations.
problem Understanding the behavior of neural networks with different activation scaling methods.
method Deriving differential equation-based asymptotic characterizations for shaped and unshaped neural networks.
result Two types of unshaped networks converge to the same infinite-depth-and-width limit at initialization.
Study finds Twitter activity correlates with stock volatility but not sentiment.
problem Understanding the impact of social media on stock market dynamics.
method Collected and analyzed tweets from Twitter and Reddit, examining their sentiment and correlation with stock volatility.
result Twitter activity correlates with stock volatility but not sentiment.
We propose a general interpretation for long-range correlation effects in the activity and volatility of financial markets. This interpretation is based on the fact that the choice between `active' and `inactive' strategies is subordinated to random-walk like processes. We numerically demonstrate our scenario in the fr…
There are non-vanishing price responses across different stocks in correlated financial markets. We further study this issue by performing different averages, which identify active and passive cross-responses. The two average cross-responses show different characteristic dependences on the time lag. The passive cross-r…
Bayesian brain computes without noise, using correlated activity.
problem Trial-to-trial variability in brain activity is not noise but probabilistic encoding.
method Analytical study of correlated neural activity and synaptic plasticity.
result Deterministic spiking networks can perform Bayesian inference without noise.
The paper establishes principles for initializing and designing GNNs with ReLU activations to avoid oversmoothing and correlation collapse.
problem Oversmoothing and correlation collapse in deep ReLU GNNs.
method The paper derives and validates three principles for initialization and architecture selection in finite width graph neural networks with ReLU activations.
result Correct initialization, residual aggregation operators, and residual connections significantly improve early training dynamics in deep ReLU GNNs.
Complex systems are typically represented by large ensembles of observations. Correlation matrices provide an efficient formal framework to extract information from such multivariate ensembles and identify in a quantifiable way patterns of activity that are reproducible with statistically significant frequency compared…
The paper validates the intensity of use model for Iran's steel consumption using economic activity indexes.
problem Validating the intensity of use model for Iran's steel consumption.
method Used support vector machines and economic activity indexes to model and predict steel consumption.
result Iran's steel consumption is strongly correlated with its economic activity.
Study finds stock selection ability of Chinese mutual funds is better than asset allocation ability.
problem Evaluating the performance of actively managed mutual funds in China.
method Developed performance measures for asset allocation and selection using holding-based models and compared them with Fama-French and Treynor-Mazuy models.
result Stock selection ability from holding-based models is positively correlated with Fama-French model, while industry allocation is positively correlated with Treynor-Mazuy model.
Neurons in cortical circuits exhibit coordinated spiking activity, and can produce correlated synchronous spikes during behavior and cognition. We recently developed a method for estimating the dynamics of correlated ensemble activity by combining a model of simultaneous neuronal interactions (e.g., a spin-glass model)…
Batch Active Learning uses derivative information for Gaussian Process regression.
problem Efficiently selecting data batches in Gaussian Process regression models.
method Proposes using the predictive covariance matrix to select data batches, exploiting full correlation.
result Demonstrates the effectiveness of incorporating derivative information across diverse applications.
Equity activity is an essential topic for financial market studies. To explore its statistical regularities, we comprehensively examine the trading value, a measure of the equity activity, of the 3314 most-traded stocks in the U.S. equity market and find that (i) the trading values follow a log-normal distribution; (ii…
Study insurance pricing under correlation ambiguity without increasing prices or reducing utility.
problem Understanding the dependence structure between insurance and financial risks.
method Dynamic equilibrium analysis of insurance pricing with worst-case beliefs.
result Correlation ambiguity does not necessarily increase insurance prices or reduce insurers' utility.
We analyse the dependence of stock return cross-correlations on the sampling frequency of the data known as the Epps effect: For high resolution data the cross-correlations are significantly smaller than their asymptotic value as observed on daily data. The former description implies that changing trading frequency sho…
Many modern databases include personal and sensitive correlated data, such as private information on users connected together in a social network, and measurements of physical activity of single subjects across time. However, differential privacy, the current gold standard in data privacy, does not adequately address p…
A novel circuit motif uses sister cells for inference with correlated priors.
problem Structured priors in neural systems pose architectural challenges.
method Proposes a novel circuit motif using sister cells to implement correlated priors without direct interactions.
result Demonstrates the efficacy of correlated priors for inference in noisy environments.
Proposes Hebbian-descent for neural network learning, addressing Hebbian and gradient descent issues.
problem Learning issues with correlated data and vanishing error term in gradient descent.
method Integrates Hebbian and gradient descent principles without activation function derivatives, centering neural activities.
result Biologically plausible, convergent, and effective in online learning with correlated data.
GRIP2 improves deep learning feature selection robustness in correlated and noisy data.
problem Identifying predictive features in correlated and noisy data.
method Integrates first-layer feature activity over a two-dimensional regularization surface to control sparsity and geometry, using efficient block-stochastic sampling.
result Demonstrates improved robustness and power in high correlation and low signal-to-noise ratio regimes.
Infinite CNNs lose spatial correlations, but can be restored by correlated weights.
problem Infinite CNNs lose spatial correlations, which are crucial for their performance.
method Introduced correlated weights to restore spatial correlations in infinite CNNs.
result Optimal performance is achieved with a moderate level of weight correlation.
Emergent misalignment is influenced by training dynamics, model priors, and data.
problem Emergent misalignment in models
method Exploring training dynamics, model priors, and data
result Activation deltas before and after narrow fine-tuning correlate with their similarities when measured with the last prompt-token activations.
Sparse activations in neural models correlate with frequent words, suggesting sparsity is natural.
problem Interpretability and resource efficiency in neural language models.
method Used the Taxi-Euclidean norm to measure sparsity and analyzed gradients and activations of frequent words.
result Frequent input words are associated with sparse activations, while frequent target words are associated with dispersed activations.
Changing initialization scale affects deep model generalization, leading to memorization or improved performance.
problem Understanding how initialization scale impacts deep model generalization and memorization.
method Experimental setup with varying initialization scales, analysis of activation and loss functions, and development of an alignment measure.
result Increasing initialization scale leads to memorization, and decreasing it improves generalization, depending on activation and loss functions.
We review the decomposition method of stock return cross-correlations, presented previously for studying the dependence of the correlation coefficient on the resolution of data (Epps effect). Through a toy model of random walk/Brownian motion and memoryless renewal process (i.e. Poisson point process) of observation ti…
Bayesian batch active learning approximates model parameters efficiently.
problem High label acquisition cost for large-scale supervised models.
method Sparse subset approximation using Bayesian active learning.
result Efficient active learning at scale with diverse batches.
Study on MC dropout in wide neural networks and its convergence to Gaussian processes.
problem Understanding the behavior of Monte Carlo dropout in wide neural networks.
method Rigorously studied the limiting distribution of wide untrained NNs under dropout, proving convergence to Gaussian processes. Investigated correlations and non-Gaussian behavior in finite width NNs.
result Wide untrained neural networks under dropout converge to Gaussian processes for fixed sets of weights and biases.
Research uses deep learning and copulas to predict multivariate survival data.
problem Handling right-censored and correlated multivariate survival data.
method Integrates deep learning, copula functions, and survival analysis. Uses copula-based activation functions to model nonlinear dependencies.
result Enhanced prediction accuracy for multivariate survival responses.
Bayesian neural networks show good correlation between out-of-sample performance and Bayesian evidence.
problem Improving the out-of-sample performance of Bayesian neural networks.
method Numerical sampling of Bayesian posterior, ensembling over architectures, analysis of evidence vs. model size.
result Good correlation between out-of-sample performance and Bayesian evidence; ensembling improves performance.
Dropout schedules can be optimized to significantly reduce model test loss.
problem Improving model performance in neural networks.
method Developed a mean-field theory of dropout at the edge of chaos, proposing front-loaded dropout schedules.
result Front-loaded dropout schedules reduce test loss by 18-35% over constant dropout.
We introduce a new activation function using Chebyshev-Lagrange polynomials for improved neural network performance.
problem Improving data efficiency and accuracy of neural networks.
method Parameterized piece-wise polynomial activation functions based on Chebyshev nodes and Lagrangian interpolation.
result Significant improvements in model capacity and accuracy, especially in linear extrapolation.
While active learning offers potential cost savings, the actual data efficiency---the reduction in amount of labeled data needed to obtain the same error rate---observed in practice is mixed. This paper poses a basic question: when is active learning actually helpful? We provide an answer for logistic regression with t…
In this work, we propose a simple yet effective solution to the problem of connectome inference in calcium imaging data. The proposed algorithm consists of two steps. First, processing the raw signals to detect neural peak activities. Second, inferring the degree of association between neurons from partial correlation …
GPRNs accurately model stellar activity affecting RV measurements of exoplanets.
problem Stellar activity limits detection and characterisation of exoplanets.
method Gaussian Process Regression Networks (GPRNs) for joint analysis of RV data and stellar activity indicators.
result GPRNs accurately describe solar RV data, correlating with activity at separations of a few days.
The Multiplicative Error Model (Engle (2002)) for nonnegative valued processes is specified as the product of a (conditionally autoregressive) scale factor and an innovation process with nonnegative support. A multivariate extension allows for the innovations to be contemporaneously correlated. We overcome the lack of …
GRASP removes spurious correlations in fine-tuned models, improving task performance and reducing bias.
problem Fine-tuned models can latch onto spurious correlations, leading to bias and reduced generalization.
method GRASP identifies and removes spurious correlations from model weights without removing latent factors.
result GRASP significantly reduces bias and improves task performance in various fine-tuning tasks.
Fragmented exchanges arise due to speed advantages in high-activity regions.
problem Fragmentation of distributed securities exchanges due to speed advantages in high-activity regions.
method Economic model and Monte Carlo simulations of a decentralized exchange with two miner clusters.
result Speed advantage increases with infrastructure asymmetry between regions.