New method improves ensemble quality by exploring the pre-train basin more effectively.
problem Limited diversity in ensembles trained from a single pre-trained checkpoint.
method Proposed StarSSE modification of Snapshot Ensembles for transfer learning.
result Stronger ensembles and uniform model soups achieved.
The paper investigates what enables successful transfer learning and separates feature reuse from data statistics.
problem Understanding what enables successful transfer learning and identifying the responsible parts of the network.
method Analyzes transfer learning on block-shuffled images to distinguish feature reuse from data statistics.
result Some benefit of transfer learning comes from learning low-level statistics of data, not just feature reuse.
New method detects metastable basins in high dimensions using trajectory sampling.
problem Identifying distinct basins in high-dimensional Markov processes.
method Discriminative approach based on marginal trajectory distribution comparison.
result Bayes-optimal classifier achieves high accuracy distinguishing between basins.
SLT reveals how grokking occurs via basin selection in training.
problem Understanding grokking in machine learning models.
method Singular Learning Theory (SLT) to analyze the loss landscape and local learning coefficient (LLC).
result LLC ranks basins by statistical preference, leading to grokking.
Proves properties of neural network basins of attraction and their expressiveness.
problem Characterize the properties of basins of attraction in neural networks.
method Analyzes width-bounded neural networks, proving properties of basins of attraction.
result Boundedness and path-connectedness of basins of attraction under certain conditions.
Wide networks are often believed to have a nice optimization landscape, but what rigorous results can we prove? To understand the benefit of width, it is important to identify the difference between wide and narrow networks. In this work, we prove that from narrow to wide networks, there is a phase transition from havi…
Quantization-aware training can recover accuracy lost by post-training quantization.
problem Post-training quantization (PTQ) can fail sharply at aggressive bitwidths.
method A unified geometric framework that explains PTQ failure and QAT recovery.
result QAT has a useful bias that steers iterates back into the basin.
Straight lines are a basin of attraction for the elastic flow at least to level 1.9615π.
problem Understanding the basin of attraction for the free boundary free elastic flow.
method Steepest descent gradient flow for elastic energy, numerical evidence.
result Straight lines have a basin of attraction at least to level 1.9615π.
Proposes a method to sample from flat basins of posterior distributions in Bayesian deep learning.
problem Sampling from multi-modal posterior distributions leads to overfitting due to trapping in bad modes.
method Introduces an auxiliary guiding variable to bias MCMC sampling towards flat basins of the energy landscape.
result The method converges faster and outperforms existing methods in sampling from flat basins of the posterior.
Study characterizes spike deconvolution basin for noisy data.
problem Recover spike locations from noisy convolution with PSF across multiple snapshots.
method Variable-projection formulation, explicit basin of convexity characterization, local convergence guarantees.
result Consistent estimator within basin of convexity under stochastic noise, complementary error bound under adversarial noise.
Neural networks' optimization dynamics are confined to a single basin despite connected basins in the loss landscape.
problem Neural networks' optimization dynamics are confined to a single basin despite connected basins in the loss landscape.
method Identifying entropic barriers arising from the interplay between curvature variations along low-loss paths and noise in optimization dynamics.
result Curvature-induced entropic forces bias noisy dynamics back toward the endpoints, explaining the confinement and connectivity of solutions.
Proposes a probabilistic model to improve hydrology predictions and trust.
problem Noisy or missing basin characteristics impact streamflow prediction.
method Probabilistic inverse modeling framework to reconstruct basin characteristics.
result 6% improvement in R2 for streamflow prediction, 17% reduction in uncertainty. This research explains why SGD generalizes better than ADAM in deep learning.
problem Understanding the generalization gap between SGD and ADAM in deep learning.
method Analyzing local convergence behaviors through Levy-driven stochastic differential equations (SDEs).
result SGD is more locally unstable and better escapes from sharp minima to flatter ones, leading to better generalization.
Study of separatrix configurations in holomorphic flows with real time.
problem Characterize separatrices in holomorphic flows with real-valued time.
method Establish continuity of transit times, classify path components, prove blow-up scenarios.
result Separatrices of different types of equilibria exhibit blow-up in finite time.
Flooding is a destructive and dangerous hazard and climate change appears to be increasing the frequency of catastrophic flooding events around the world. Physics-based flood models are costly to calibrate and are rarely generalizable across different river basins, as model outputs are sensitive to site-specific parame…
Quantification of the stationary points and the associated basins of attraction of neural network loss surfaces is an important step towards a better understanding of neural network loss surfaces at large. This work proposes a novel method to visualise basins of attraction together with the associated stationary points…
Study global geometry of dynamical systems with entire vector fields.
problem Understanding the global structure of equilibria and their basins.
method Step-by-step analysis of basins of centers, nodes, and foci; introduction of global elliptic sectors.
result Characterization of heteroclinic regions connecting equilibria.
Regional rainfall-runoff modeling is an old but still mostly out-standing problem in Hydrological Sciences. The problem currently is that traditional hydrological models degrade significantly in performance when calibrated for multiple basins together instead of for a single basin alone. In this paper, we propose a nov…
We study codimension one foliations with singularities defined locally by Bott-Morse functions on closed oriented manifolds. We carry to this setting the classical concepts of holonomy of invariant sets and stability, and prove a stability theorem in the spirit of the local stability theorem of Reeb. This yields, among…
New method finds basins of attraction without needing system models.
problem Determining basins of attraction (BoA) for nonlinear systems without prior knowledge.
method Hybrid Active Learning (HAL) method combining AST, AL, and DBS.
result Efficiently finds and labels boundary of BoA without model knowledge.
Researchers use Gaussian processes with non-stationary kernels to model precipitation patterns in the Upper Indus Basin.
problem Uncertainty in precipitation patterns in the Upper Indus Basin, Himalayas.
method Proposes Gaussian processes with structured non-stationary kernels to model precipitation patterns, accounting for spatial variation with a latent Gaussian process.
result The proposed model adapts to varying precipitation patterns across distinct topography and outperforms stationary models in ablation experiments.
SGD transitions between maxima and minima with varying time scales.
problem Understanding SGD's behavior near critical points in noisy landscapes.
method Analyzing SGD convergence and escape dynamics in 1D landscapes with infinite- and finite-variance noise.
result SGD reliably moves to the basin's minimum unless close to a local maximum, where it can linger.
Study SRB measures for Anosov actions on manifolds.
problem Characterize SRB measures for Anosov actions.
method Use Ruelle-Taylor resonances and properties of Sinai-Ruelle-Bowen measures.
result SRB measures have properties like smooth disintegrations, positive basins, and are unique under certain conditions.
Multifractality in time series arises from temporal correlations, not just fat tails.
problem Understanding the origin of multifractality in time series data.
method Mathematical arguments and numerical simulations using MFDFA approach.
result Genuine multifractality requires temporal correlations, not just fat tails.
Biological neurons learn tensor decompositions of higher-order correlations using nonlinear Hebbian plasticity.
problem Learning higher-order correlations in biological neurons.
method Introduce and study generalized nonlinear Hebbian learning rules.
result Neurons can learn tensor eigenvectors of higher-order input correlation tensors.
WASH trains ensembles with shuffled weights to improve accuracy and reduce communication.
problem Training ensembles for weight averaging leads to models converging to different loss basins.
method WASH randomly shuffles a small percentage of weights during training to keep models within the same basin.
result WASH achieves state-of-the-art image classification accuracy with lower communication costs.
Recently, a general method for analyzing the statistical accuracy of the EM algorithm has been developed and applied to some simple latent variable models [Balakrishnan et al. 2016]. In that method, the basin of attraction for valid initialization is required to be a ball around the truth. Using Stein's Lemma, we exten…
Paper proposes a new method for population-wise matching of sulcal graphs.
problem Challenges in matching cortical fold variations across individuals.
method Population-wise multi-graph matching of sulcal graphs.
result Effectiveness of multi-graph matching in obtaining consistent labeling of sulcal basins.
Framework analyzes neural network dynamics for better understanding and optimization.
problem Understanding the fundamental mechanisms of deep neural networks.
method Dynamical systems theory, transformation units, attraction basins.
result Different transformation modes lead to distinct learning phases and network performance.
DeepWeightFlow generates diverse neural network weights efficiently.
problem Generating complete neural network weights efficiently and accurately.
method Flow Matching in weight space with Git Re-Basin and TransFusion.
result DeepWeightFlow generates high-accuracy neural networks without fine-tuning.
We introduce a mathematical model on the dynamics of demand and supply incorporating collectability and saturation factors. Our analysis shows that when the fluctuation of the determinants of demand and supply is strong enough, there is chaos in the demand-supply dynamics. Our numerical simulation shows that such a cha…
Study of holomorphic correspondences combining entire maps and Fuchsian groups.
problem Understanding dynamics of entire maps and their interactions with Fuchsian groups.
method Systematic study of (∞:∞) holomorphic correspondences arising from conformal combinations of transcendental entire maps and Fuchsian groups. result The resulting correspondence is the composition of a Möbius involution and the deleted covering correspondence of a meromorphic function with a simple pole.
We consider compressed sensing formulated as a minimization problem of nonconvex sparse penalties, Smoothly Clipped Absolute deviation (SCAD) and Minimax Concave Penalty (MCP). The nonconvexity of these penalties is controlled by nonconvexity parameters, and L1 penalty is contained as a limit with respect to these para…
An image pattern can be represented by a probability distribution whose density is concentrated on different low-dimensional subspaces in the high-dimensional image space. Such probability densities have an astronomical number of local modes corresponding to typical pattern appearances. Related groups of modes can join…
Recent observations with varied schedules and types (moving average, snapshot, or regularly spaced) can help to improve streamflow forecasts, but it is challenging to integrate them effectively. Based on a long short-term memory (LSTM) streamflow model, we tested multiple versions of a flexible procedure we call data i…
HydroNets use river structure to improve hydrologic predictions.
problem Scalable and accurate hydrologic models are needed for climate change impacts.
method HydroNets are deep neural networks that incorporate river network structure.
result HydroNets improve predictions with fewer data, especially at longer horizons.
Deep learning, in the form of artificial neural networks, has achieved remarkable practical success in recent years, for a variety of difficult machine learning applications. However, a theoretical explanation for this remains a major open problem, since training neural networks involves optimizing a highly non-convex …
GP-TS optimizes TLM pre-training hyperparameters efficiently.
problem Resource inefficiency in TLM pre-training.
method Bayesian optimization with Thompson sampling and Gaussian process.
result GP-TS achieves lower MLM loss in fewer epochs.
RAMBO optimizes multi-regime problems by discovering and modeling distinct energy basins.
problem Multi-regime problems in molecular conformation and drug discovery.
method Dirichlet Process Mixture of Gaussian Processes with adaptive hyperparameters and concentration parameters.
result Consistent improvements over state-of-the-art on multi-regime objectives.
A new associative memory uses Sinkhorn divergence for efficient pattern retrieval.
problem Efficiently retrieving patterns from large datasets of weighted point clouds.
method Derived retrieval dynamics as a SHK gradient flow, discretized for a deterministic algorithm.
result Proved basin invariance, geometric convergence, and robust recovery from perturbations.
The paper explains how continuous language models can produce discrete, interpretable meanings.
problem Semantic collapse in continuous systems of large language models.
method Formalizing large language models as Continuous State Machines (CSMs) and analyzing the associated transfer operator.
result The leading eigenfunctions of the transfer operator induce a finite number of invariant meaning basins, explaining how continuous computation can produce discrete, interpretable semantics.
Self-training outperforms pre-training on COCO object detection and segmentation datasets.
problem The effectiveness of pre-training in improving object detection and segmentation models is limited.
method Investigated self-training as an alternative method to utilize additional data.
result Self-training consistently improves model performance across various dataset sizes and data augmentation levels.
Recent work has demonstrated the effectiveness of gradient descent for directly recovering the factors of low-rank matrices from random linear measurements in a globally convergent manner when initialized properly. However, the performance of existing algorithms is highly sensitive in the presence of outliers that may …
He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. …
New framework explains how larger pre-trained models reduce downstream learning sample complexity.
problem Understanding why larger pre-trained models reduce sample complexity in downstream tasks.
method Introducing a novel framework called Caulking inspired by PEFT methods.
result Improved pre-trained models provably decrease downstream task sample complexity.
Pre-training on different modalities improves Transformer performance in offline reinforcement learning.
problem Improving Transformer-based models for offline reinforcement learning tasks.
method Empirical investigation of pre-training on language and vision data, analysis of internal representations, and fine-tuning with no context.
result Pre-trained Transformers with language data can utilize context-like information to solve reinforcement learning tasks efficiently.
The study finds a theoretical bound for pre-training iterations needed for pruning to yield good subnetwork performance.
problem Discovering efficient subnetworks within pre-trained dense networks.
method Mathematical analysis of a two-layer, fully-connected network, validating with a multi-layer perceptron trained on MNIST.
result A logarithmically dependent threshold on dataset size for successful pruning.
Unsupervised pre-training improves model generalization, but lacks theoretical understanding.
problem Lack of theoretical understanding of unsupervised pre-training's impact on model generalization.
method Introduces a novel theoretical framework to analyze and enhance generalization.
result Enhances understanding of unsupervised pre-training and fine-tuning, proposing a new regularization method.