Labels distilled from images improve model training efficiency and flexibility.
problem Creating synthetic labels for a small set of real images to train models effectively.
method Introduce a more robust and flexible meta-learning algorithm for distillation and an effective first-order strategy based on convex optimization layers.
result Label distillation leads to improved results and greater flexibility in neural architectures.
We present any-precision deep neural networks (DNNs), which are trained with a new method that allows the learned DNNs to be flexible in numerical precision during inference. The same model in runtime can be flexibly and directly set to different bit-widths, by truncating the least significant bits, to support dynamic …
PFNs4BO uses neural processes for flexible Bayesian Optimization.
problem Efficient surrogate modeling for Bayesian Optimization.
method In-context learning of PFNs to approximate posterior predictive distribution.
result PFNs outperform traditional GP, BNN in BO tasks.
Fiber simplifies RL and population-based methods for distributed training.
problem Challenges in RL and population-based methods, including frequent interaction with simulations and dynamic scaling.
method Introducing Fiber, a scalable distributed computing framework.
result Significantly expands accessibility of large-scale parallel computation.
Flexible per-class regularization improves binary classifiers.
problem Improving binary classifiers by addressing outliers and class imbalance.
method Graph-based adaptive regularization with flexible per-class thresholds.
result Flexible thresholds improve classifier performance and address class imbalance.
ACE models allow flexible conditioning and prediction of latent variables.
problem Lack of flexibility in conditioning and prediction of latent variables in probabilistic models.
method Introduces Amortized Conditioning Engine (ACE) that explicitly represents latent variables and allows runtime conditioning and prediction.
result ACE models outperform existing methods in diverse tasks like image completion, classification, Bayesian optimization, and simulation-based inference.
A new method combines experts' opinions to train regression models with noisy labels.
problem Training regression models with noisy labels from multiple experts.
method Estimate each labeler's expertise and combine opinions using learned weights.
result Empirically outperforms existing techniques on simulated and real data.
Flexible tree ensemble learning framework supports arbitrary loss functions and multi-task learning.
problem Limited modeling capabilities of existing tree ensemble learning toolkits.
method Differentiable tree ensembles with tensor-based formulation for efficient training.
result Our framework leads to 100x more compact and 23% more expressive tree ensembles.
Neural networks fit fewer samples than their parameters suggest in practice.
problem Understanding the practical limitations of neural network flexibility.
method Examination of neural network optimization, parameter efficiency, and loss surfaces.
result Neural networks can only fit training sets with significantly fewer samples than their parameters suggest.
Simformer uses transformer models to perform flexible Bayesian inference.
problem Current simulation-based inference methods are inflexible and require fixed priors.
method Trains a probabilistic diffusion model with transformer architectures.
result Outperforms state-of-the-art methods on various benchmarks.
Gradient Boosted Normalizing Flows improve flexibility of NFs without increasing complexity.
problem Improving flexibility of normalizing flows without increasing complexity.
method Gradient Boosting applied to normalizing flows to create a mixture model structure.
result GBNFs outperform non-boosted NFs and produce better results with simpler components.
Conditional Text Generation has drawn much attention as a topic of Natural Language Generation (NLG) which provides the possibility for humans to control the properties of generated contents. Current conditional generation models cannot handle emerging conditions due to their joint end-to-end learning fashion. When a n…
Existing Markov Chain Monte Carlo (MCMC) methods are either based on general-purpose and domain-agnostic schemes which can lead to slow convergence, or hand-crafting of problem-specific proposals by an expert. We propose A-NICE-MC, a novel method to train flexible parametric Markov chain kernels to produce samples with…
GNet uses Gaussian processes for scalable, flexible neural networks.
problem Large-scale predictive modeling with high computational and storage costs.
method GNet employs Gaussian processes with nonparametric activation functions and a fast algorithm for training and predictions.
result GNet achieves competitive performance across various test problems, including nonlinear function prediction and real-world data regression.
GNet uses Gaussian processes for scalable, flexible neural networks.
problem Large-scale predictive modeling with high computational and storage costs.
method GNet employs Gaussian processes with nonparametric activation functions and a fast algorithm for efficient training and predictions.
result GNet achieves competitive performance across various test problems, including nonlinear function prediction and real-world data regression.
EBMs are flexible but hard to train; this paper explains methods.
problem Training Energy-Based Models is difficult due to the unknown normalizing constant.
method Explains MCMC, SM, and NCE for training EBMs, highlighting connections.
result Provides a friendly introduction to modern EBM training methods.
Proposes flexible auto-encoders for varying data dimensions.
problem Fixed latent dimensions limit data flexibility.
method Stochastic bottleneck with weighted dropouts.
result Seamless variable dimensionality reduction with high performance.
A key advance in learning generative models is the use of amortized inference distributions that are jointly trained with the models. We find that existing training objectives for variational autoencoders can lead to inaccurate amortized inference distributions and, in some cases, improving the objective provably degra…
We consider the problem of training generative models with deep neural networks as generators, i.e. to map latent codes to data points. Whereas the dominant paradigm combines simple priors over codes with complex deterministic models, we argue that it might be advantageous to use more flexible code distributions. We de…
Paper adapts diffusion sampler training for faster convergence and better sampling.
problem Training limitations in diffusion samplers.
method Decouples generation and destruction variances, learns both as unconstrained Gaussians.
result Training both processes leads to faster convergence and improved sampling quality.
BI-EqNO improves Bayesian inference with flexible neural operators.
problem Inaccurate estimation of marginal likelihoods in approximate Bayesian methods.
method Equivariant neural operator framework for generalized approximate Bayesian inference.
result BI-EqNO enhances both deterministic and stochastic approaches to Bayesian inference.
Multi-task learning (MTL) is a common paradigm that seeks to improve the generalization performance of task learning by training related tasks simultaneously. However, it is still a challenging problem to search the flexible and accurate architecture that can be shared among multiple tasks. In this paper, we propose a …
Flexible device participation improves federated learning convergence.
problem Strict device participation limits federated learning reach.
method Analytical results and new aggregation scheme for flexible participation.
result Convergence improved with flexible device participation.
From only positive (P) and unlabeled (U) data, a binary classifier could be trained with PU learning, in which the state of the art is unbiased PU learning. However, if its model is very flexible, empirical risks on training data will go negative, and we will suffer from serious overfitting. In this paper, we propose a…
This work improves policy-based training by proposing an evaluation balance objective for GFlowNets.
problem Reliable estimation of policy divergence under directed acyclic graphs remains challenging.
method Proposes an evaluation balance objective over partial episodes to measure policy divergence and improve policy-based training reliability.
result Evaluation balance strengthens policy-based training reliability and broadens its flexibility.
Proposes a variational autoencoder for long-term customer revenue forecasting.
problem Predicting long-term customer revenue from sparse and irregular transaction data.
method Variational Autoencoder (VAE) with flexible latent representation.
result Improves upon latest benchmarks in multiple real-world datasets.
EBPs model exchangeable data with flexible distributions.
problem Current energy-based models restrict set cardinality and limited distribution forms.
method Introduced Energy-Based Processes (EBPs) that extend energy models to exchangeable data with neural network parameterizations.
result EBPs can express more flexible distributions over sets without cardinality restrictions.
Proposes a new method for continual learning in neural networks.
problem Challenges in applying sequential Bayesian inference to neural networks.
method Sequential function-space variational inference.
result Neural networks trained with the proposed method achieve better predictive accuracy.
We present a training system, which can provably defend significantly larger neural networks than previously possible, including ResNet-34 and DenseNet-100. Our approach is based on differentiable abstract interpretation and introduces two novel concepts: (i) abstract layers for fine-tuning the precision and scalabilit…
Deep reinforcement learning for high dimensional, hierarchical control tasks usually requires the use of complex neural networks as functional approximators, which can lead to inefficiency, instability and even divergence in the training process. Here, we introduce stacked deep Q learning (SDQL), a flexible modularized…
We introduce the vine copula autoencoder (VCAE), a flexible generative model for high-dimensional distributions built in a straightforward three-step procedure. First, an autoencoder (AE) compresses the data into a lower dimensional representation. Second, the multivariate distribution of the encoded data is estimated …
In a range of fields including the geosciences, molecular biology, robotics and computer vision, one encounters problems that involve random variables on manifolds. Currently, there is a lack of flexible probabilistic models on manifolds that are fast and easy to train. We define an extremely flexible class of exponent…
Proposes a method to model financial returns with extreme shocks using flexible tail transformations.
problem Capturing extreme shocks in financial return data.
method Introduces a transformation layer in normalizing flows to model heavy-tailed distributions.
result Trained models can generate synthetic sets of extreme returns.
Flexible framework compresses models using LC algorithm.
problem Efficiently compressing neural networks for resource constraints.
method Decouples learning and compression steps with alternating L and C phases.
result Compressed models maintain performance and accuracy.
Deep Gaussian processes (DGPs) are multi-layer hierarchical generalisations of Gaussian processes (GPs) and are formally equivalent to neural networks with multiple, infinitely wide hidden layers. DGPs are probabilistic and non-parametric and as such are arguably more flexible, have a greater capacity to generalise, an…
Support Vector Data Description (SVDD) is a popular outlier detection technique which constructs a flexible description of the input data. SVDD computation time is high for large training datasets which limits its use in big-data process-monitoring applications. We propose a new iterative sampling-based method for SVDD…
DRMMs enable flexible conditional sampling for interactive machine learning.
problem Limited flexibility in conditional sampling for deep generative models.
method Proposes Deep Residual Mixture Models (DRMMs) that allow flexible conditional sampling.
result DRMMs enable sampling with arbitrary combinations of conditioning variables and priors.
Federated learning improves by training central model with client model outputs.
problem Direct averaging of client models is limited in FL.
method Ensemble distillation for model fusion.
result Central model trained faster with fewer communication rounds.
Neural networks have proven to be extremely powerful tools for modern artificial intelligence applications, but computational and storage complexity remain limiting factors. This paper presents two compatible contributions towards reducing the time, energy, computational, and storage complexities associated with multil…
RNE provides a flexible framework for diffusion models, enabling inference-time control and energy-based training.
problem Insufficient knowledge of marginal densities in diffusion models.
method Introduces Radon-Nikodym Estimator (RNE) to reveal the connection between marginal densities and transition kernels.
result RNE delivers strong results in inference-time control and energy-based diffusion training.
Operating deep neural networks (DNNs) on devices with limited resources requires the reduction of their memory as well as computational footprint. Popular reduction methods are network quantization or pruning, which either reduce the word length of the network parameters or remove weights from the network if they are n…
Paper improves training physics-informed neural networks with model ensembles.
problem Training physics-informed neural networks (PINNs) is difficult due to convergence to wrong solutions.
method Proposes training an ensemble of PINNs, using ensemble agreement to expand the solution interval.
result Algorithm stabilizes PINN training and yields competitive performance.
Extended LSTMs improve volatility prediction by 20%.
problem Predicting asset price volatility with long memory.
method Extended LSTMs with multiple flexible timescales.
result Extended LSTMs outperform rough volatility predictions by 20%.
RegFlow models future states with flexible probability distributions.
problem Predicting future states under complex, non-deterministic scenarios.
method Hypernetwork architecture and continuous normalizing flow model.
result RegFlow achieves state-of-the-art results on benchmark datasets.
New method SF-AdamW trains large models without decay phases or memory overhead.
problem Inadequate fixed compute budgets for large-scale training.
method Schedule-Free (SF) method revisited and refined.
result SF-AdamW effectively navigates loss landscape without decay phases or memory overhead.
Calcium imaging permits optical measurement of neural activity. Since intracellular calcium concentration is an indirect measurement of neural activity, computational tools are necessary to infer the true underlying spiking activity from fluorescence measurements. Bayesian model inversion can be used to solve this prob…
A new model uses Toeplitz matrices to analyze time-series data transitions.
problem Analyzing transitions in time-series data from nonautonomous systems.
method Deep Koopman-layered models with learnable Toeplitz matrices, leveraging Toeplitz matrices' universal property.
result The model demonstrates universality and generalization, outperforming existing methods.
PSI models and infers feature attributions efficiently and accurately.
problem Modeling and inferring feature attributions in flexible predictive models.
method Probabilistic Shapley inference (PSI) framework using latent random variables and a masking-based neural network architecture.
result PSI learns feature attribution distributions centered at Shapley values, revealing meaningful uncertainty.