Geodesics connect model modes in neural network loss landscapes.
problem Connecting modes in neural network loss landscapes.
method Reframed mode connectivity in Information Geometry, hypothesized geodesics as mode-connecting paths, proposed algorithm to approximate geodesics.
result Geodesics achieve mode connectivity in neural networks.
Proposes using mode connectivity to improve adversarial robustness of neural networks.
problem Improving adversarial robustness of deep neural networks.
method Employing mode connectivity in loss landscapes to study adversarial robustness and propose methods for improvement.
result Path connection learned using limited bonafide data can effectively mitigate adversarial effects while maintaining original accuracy.
Stable GFlowNets prevent loss spikes and mode collapse in training.
problem Unstable training of GFlowNets leading to loss spikes and mode collapse.
method Assessed sensitivity of GFlowNet objectives, derived loss-to-TV bounds, and proposed Stable GFlowNets.
result Stable GFlowNets improve training behavior and distributional fidelity.
A new GAN model α-GAN with tunable loss function addresses gradient vanishing and mode collapse issues.
problem Addressing vanishing gradients and mode collapse in GANs.
method Introduced a tunable GAN α-GAN using a supervised α-loss function. result Holistic understanding of α-GAN related to Arimoto divergence and convergence properties. Proposes neuron alignment to optimize mode connectivity in neural networks.
problem Understanding and optimizing mode connectivity in deep neural networks.
method Introduces neuron alignment to approximate optimal weight permutations and improve mode connectivity.
result Neuron alignment significantly alleviates robust loss barriers and improves model robustness and accuracy.
Paper discovers simplicial complexes connecting trained models for improved ensembling.
problem Improving robustness and accuracy of deep learning ensembles.
method Identifies mode-connecting simplicial complexes on loss surfaces.
result Efficiently builds simplicial complexes for ensembling, outperforming independent ensembles.
Intervention loss stabilizes GAN training and reduces mode collapse.
problem Stabilizing GAN training and mitigating mode collapse.
method Intervention loss regularization term introduced into GAN objective.
result Improved GAN training stability and reduced mode collapse.
Prb-GAN uses dropout and variational inference to improve GAN performance.
problem GANs struggle with mode loss and training instability.
method Introduces Prb-GANs with dropout and variational inference for parameter distribution.
result Improves GAN performance through dropout and variational inference.
The loss functions of deep neural networks are complex and their geometric properties are not well understood. We show that the optima of these complex loss functions are in fact connected by simple curves over which training and test accuracy are nearly constant. We introduce a training procedure to discover these hig…
Proposes hinge-Wasserstein to improve uncertainty estimation in regression tasks.
problem Estimating multimodal aleatoric uncertainty in regression tasks from images.
method Regression-by-classification paradigm with hinge-Wasserstein loss.
result Hinge-Wasserstein loss improves uncertainty estimation on challenging tasks.
Deep generative models provide powerful tools for distributions over complicated manifolds, such as those of natural images. But many of these methods, including generative adversarial networks (GANs), can be difficult to train, in part because they are prone to mode collapse, which means that they characterize only a …
Generative Adversarial Networks are known for their high quality outputs and versatility. However, they also suffer the mode collapse in their output data distribution. There have been many efforts to revamp GANs model and reduce mode collapse. This paper focuses on two of these models, PacGAN and VEEGAN. This paper ex…
LoRA-Curve connects independent LoRA optima through continuous low-loss valleys, improving Bayesian model averaging.
problem Challenges in estimating epistemic uncertainty in LoRA-based Bayesian inference.
method Introduces LoRA-Curve, a segmented Bézier curve parameterization in the LoRA space, with free and anchored configurations.
result Empirically shows that connecting independent LoRA optima through continuous low-loss valleys improves mutual information of the predictive distribution.
LMC loss barrier decreases to zero with large network width.
problem Understanding the LMC phenomenon in neural networks.
method Fine-grained analysis of LMC for two-layer ReLU networks.
result LMC loss barrier decreases to zero at a rate of O(m^-1/2) for large network width.
Proposes IIB for domain generalization, overcoming failure modes of IRM.
problem Domain generalization with nonlinear classifiers and pseudo-invariant features.
method Invariant Information Bottleneck (IIB) using mutual information and variational formulation.
result Significantly outperforms IRM on synthetic datasets and real-world benchmarks.
Improved training of RBMs using mode-assisted gradient updates.
problem Difficulty in training RBMs due to hard-to-compute gradients.
method Combining standard gradient updates with mode-assisted off-gradient direction.
result Mode training improves training speed and stability, reduces KL divergence.
Generative adversarial networks (GANs) often suffer from unpredictable mode-collapsing during training. We study the issue of mode collapse of Boundary Equilibrium Generative Adversarial Network (BEGAN), which is one of the state-of-the-art generative models. Despite its potential of generating high-quality images, we …
Proposes a new loss function for robust learning.
problem Creating a robust loss function for machine learning.
method Extended pseudo Huber loss with log-exp transform and logistic function.
result Linear convergence algorithm for minimizer finding.
Diffusion models accurately recover mixture weights from generated samples despite score function insensitivity.
problem Score-based generative models often fail to learn correct relative mode amplitudes (mixture weights) from generated samples.
method Relate diffusion score matching (DSM) loss to mixture weight estimation error, define diffusion score sensitivity index (DSSI), and prove its governing role in mixture weight recovery.
result Generated samples can accurately recover mixture weights from the DSM loss, even when the target score is insensitive to mixture weights.
Mode connectivity is a recently introduced frame- work that empirically establishes the connected- ness of minima by finding a high accuracy curve between two independently trained models. To investigate the limits of this setup, we examine the efficacy of this technique in extreme cases where the input models are trai…
The paper proves inequalities for zero mode equations on spin manifolds.
problem Sharp inequalities for zero mode equations on spin manifolds.
method Simple proof and classification of equality cases.
result Equality holds if and only if the solution is a Killing spinor and the manifold is Sasaki-Einstein.
Deep ensembles improve model accuracy and robustness, but their theoretical underpinnings are not fully understood.
problem Understanding why deep ensembles work well in practice despite theoretical limitations.
method Investigating the loss landscape of neural networks and exploring the diversity of functions in function space.
result Random initializations explore diverse modes in function space, while ensembles along an optimization trajectory cluster within a single mode.
Cautious Weight Decay modifies weight decay for better optimization.
problem Improving optimization in deep learning models.
method Applies weight decay selectively based on parameter sign alignment.
result Consistently improves model performance across various tasks and scales.
We improve generative models for heavy-tailed multivariate data using an invariant statistical loss.
problem Traditional generative models struggle with heavy-tailed and multivariate data, leading to unstable training and mode dropping.
method We extend the invariant statistical loss method to handle heavy-tailed and multivariate data using a Pareto-ISL generator trained with input noise from a generalised Pareto distribution.
result Pareto-ISL accurately models the tails of heavy-tailed distributions while capturing central characteristics.
New metrics improve scRNA-seq perturbation modeling by reducing mode collapse.
problem Outperformed by simple mean prediction in scRNA-seq perturbation modeling.
method Introduce DEG-aware metrics (WMSE, Rw2(Δ)) and negative/positive baselines. result WMSE loss function reduces mode collapse and improves model performance.
The paper explores trading off consistency and dimensionality in convex surrogates for multiclass classification.
problem Designing consistent surrogate losses for multiclass classification with high-dimensional outcomes.
method Investigates embedding outcomes into convex polytopes and examining consistency under low-noise assumptions.
result Consistency can be achieved with less than n−1 dimensions, but hallucination occurs for some distributions. Solves GAN mode collapse by assigning minibatches to multiple discriminators.
problem Mode collapse in GANs where models generate similar samples.
method Multiple discriminators, microbatching, and changing tasks.
result Promotes sample diversity in generated sets.
Adaptive teacher improves sample efficiency and mode coverage in sampling tasks.
problem Efficient exploration and mode coverage in sampling tasks with amortized inference.
method An adaptive training distribution (teacher) guides the training of the primary sampler (student).
result Improves sample efficiency and mode coverage across various tasks.
New method prevents mode collapse in deep SVDD for anomaly detection.
problem Mode collapse in deep SVDD due to architectural constraints.
method Two regularizers: noise injection and minibatch variance penalization.
result Regularized deep SVDD outperforms state-of-the-art methods.
Unified framework connects different neural network models.
problem Understanding the geometry of neural network loss landscapes.
method Unified framework capturing four symmetry classes.
result First discovery of low- and zero-barrier linear interpolation paths.
Koopman mode analysis applied to neural networks for training optimization.
problem Optimizing neural network training, identifying issues, and speeding up learning.
method Koopman operator analysis of neural network dynamics.
result Spectral analysis of Koopman operator aids in determining network depth, initialization quality, and training termination.
As an effective way of metric learning, triplet loss has been widely used in many deep learning tasks, including face recognition and person-ReID, leading to many states of the arts. The main innovation of triplet loss is using feature map to replace softmax in the classification task. Inspired by this concept, we prop…
A new method improves Bayesian inference for multimodal posteriors.
problem Insensitivity to well-separated modes in multimodal posteriors.
method Weighted Kernel Stein Discrepancy method.
result Significantly improved mode sensitivity compared to standard KSD-Bayes.
Generalized dual discriminator GANs improve upon traditional GANs by using two discriminators and a flexible loss function.
problem Mode collapse in GANs.
method Introducing dual discriminator α-GANs and extending the approach to arbitrary functions. result The approach reduces the optimization problem to a linear combination of an f-divergence and a reverse f-divergence. This work aims to reduce inexplicable errors in deep neural networks by obtaining class-level semantics and penalizing misclassifications.
problem Deep neural networks misclassify images, leading to inexplicable errors that can harm trust and societal impact.
method Obtain class-level semantics, propose Weighted Loss Functions (WLFs), and train classifiers with these methods.
result Trained networks have more explicable failure modes and comparable accuracy to existing methods.
While the Matrix Generalized Inverse Gaussian (MGIG) distribution arises naturally in some settings as a distribution over symmetric positive semi-definite matrices, certain key properties of the distribution and effective ways of sampling from the distribution have not been carefully studied. In this paper…
Recent work on mode connectivity in the loss landscape of deep neural networks has demonstrated that the locus of (sub-)optimal weight vectors lies on continuous paths. In this work, we train a neural network that serves as a hypernetwork, mapping a latent vector into high-performance (low-loss) weight vectors, general…
A two-layer classifier improves smartphone transportation mode recognition.
problem Improving accuracy of transportation mode classification.
method Two-layer hierarchical classifier combining time and frequency domain features.
result Maximum classification accuracy of 97.02%.
Improves conditions for mode connectivity in deep neural networks.
problem Ensuring mode connectivity in deep neural networks under practical conditions.
method Exploiting feature quality and mild over-parameterization conditions.
result Connectivity of solutions found by stochastic gradient descent confirmed under new conditions.
New method trains generative models without discriminators, improving stability and accuracy.
problem Training implicit generative models with adversarial discriminators leads to instability and mode-dropping.
method Invariant statistical loss function, avoiding discriminators.
result Successfully trains generative models for various complex distributions without mode-dropping.
Paper tackles tensor decomposition for unaligned observations using RKHS and novel loss functions.
problem Tackles tensor decomposition for unaligned observations.
method Uses functions in RKHS to represent mode with unaligned observations, introduces versatile loss function, proposes optimization algorithm and stochastic gradient method.
result Demonstrates improved tensor decomposition efficiency and effectiveness with synthetic and real data.
Unsupervised domain transfer is the task of transferring or translating samples from a source distribution to a different target distribution. Current solutions unsupervised domain transfer often operate on data on which the modes of the distribution are well-matched, for instance have the same frequencies of classes b…
Paper proposes ManiF-SMC for effective approximate machine unlearning.
problem Limited unlearning effectiveness and potential to undermine original learning objectives.
method Reformulates approximate unlearning as pushing erased samples towards semantic neighbors in retained data, using a margin-based triplet loss.
result Achieves unlearning effectiveness comparable to state-of-the-art methods while operating purely in representation space.
Randomized fiber projections enhance neural network accuracy.
problem Improving neural network performance with limited resources.
method Training neural networks on randomized speckled images from multi-mode fiber projectors.
result Classification accuracy is higher with randomized fiber data than with direct images.
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.
Generative adversarial networks (GANs) are a powerful approach to unsupervised learning. They have achieved state-of-the-art performance in the image domain. However, GANs are limited in two ways. They often learn distributions with low support---a phenomenon known as mode collapse---and they do not guarantee the exist…
Simplified non-contrastive learning avoids representation collapse.
problem Training failure modes in self-supervised learning.
method Hyperdimensional computing and inductive bias.
result The approach avoids representation collapses.
CNFs model conditional densities efficiently and avoid mode collapse.
problem Capturing strong correlations between output dimensions in structured prediction tasks.
method Conditional normalizing flows (CNFs) with a likelihood-based objective.
result CNFs achieve competitive performance in super-resolution and vessel segmentation tasks.