A new cost-frugal HPO method controls training cost during optimization.
problem Ignoring training cost variation in HPO leads to inefficient hyperparameter tuning.
method Developed a randomized direct-search method with convergence and approximation guarantees.
result Proved an O ( d K ) O(\frac{\sqrt{d}}{\sqrt{K}}) O ( K d ) convergence rate and O ( d ε − 2 ) O(dε^{-2}) O ( d ε − 2 ) approximation guarantee. New model considers varying costs in learning, outperforming existing methods.
problem Learning with varying costs in machine learning models.
method Introduces ε-frugal learning that considers both known and unknown costs.
result ε-frugal learners outperform learners with known costs and random sampling.
Frugal method predicts multiple local electricity loads efficiently.
problem Day-ahead forecasting of over 1000 substations in France.
method Adaptive generalized additive models with state-space representations, combined with transfer learning.
result Reduction of computational needs and emissions with competitive accuracy.
A new optimizer saves significant shots in quantum machine learning.
problem Training variational QML algorithms is challenging due to large datasets and shot-count overhead.
method Proposed Refoqus optimizer that samples over both dataset and measurement operators.
result Refoqus can save several orders of magnitude in shot cost compared to existing methods.
Frugal Flows learn complex data and infer marginal causal effects.
problem Challenges in estimating marginal causal effects from complex data.
method Frugal Flows use normalizing flows to flexibly learn data and infer causal quantities.
result Frugal Flows can generate synthetic data that closely matches real-world data and exactly parameterize causal quantities.
Spectral regularization improves learning over combinatorial spaces with limited data.
problem Learning pseudo-Boolean functions with scarce labeled data.
method Regularizing the spectral representation of learned functions using the L_1 norm.
result Regularization allows for data-frugal learning and achieves statistically optimal generalization performance.
Frengression models causal data flexibly and faithfully.
problem Challenges in robust benchmarking and evaluation of causal inference with real-world data.
method Introduces frengression, a deep generative model for joint distribution of covariates, treatments, and outcomes.
result Frengression provides accurate estimation and flexible simulation of multivariate, time-varying data.
Sampling a fraction of pairs can match full evaluation in machine learning losses.
problem High computational cost of full pairwise loss evaluation.
method Survey sampling techniques targeting informative pairs.
result Performance close to full pairwise evaluation achieved with frugal sampling.
Proposes a method to evaluate generalizability in causal inference models.
problem Lack of formal procedures to statistically evaluate generalizability in causal inference.
method Frugal parameterization to simulate from causal benchmarks, using mean and distributional regression methods.
result Ensures more realistic evaluations of causal inference models, avoiding over-reliance on conventional metrics.
New approach selects sparse features without validation.
problem Overfitting and high storage/compute costs in AI models.
method Phase transition in probability of feature retrieval.
result Flexible approach for various models and loss functions.
Model learns brevity by exposing to easy problems, improving efficiency without explicit length penalties.
problem Excessive verbosity in step-by-step reasoning models trained with RLVR.
method Retaining and up-weighting moderately easy problems as implicit length regularizers.
result Model generates solutions that are, on average, nearly twice as short without explicit length penalties.
We present an LDA approach to entity disambiguation. Each topic is associated with a Wikipedia article and topics generate either content words or entity mentions. Training such models is challenging because of the topic and vocabulary size, both in the millions. We tackle these problems using a novel distributed infer…
The family of temporal difference (TD) methods span a spectrum from computationally frugal linear methods like TD(λ) to data efficient least squares methods. Least square methods make the best use of available data directly computing the TD solution and thus do not require tuning a typically highly sensitive learning r…
We provide a mathematical definition of fragility and antifragility as negative or positive sensitivity to a semi-measure of dispersion and volatility (a variant of negative or positive "vega") and examine the link to nonlinear effects. We integrate model error (and biases) into the fragile or antifragile context. Unli…
Paper introduces SPADE method to protect classifiers from OOD and adversarial samples.
problem Protecting classifiers from out-of-distribution and adversarial samples.
method SPADE method based on GEV model in latent space.
result Provable protection against OOD and adversarial samples.
Bayesian approach selects features for a specific target with high confidence.
problem Feature acquisition costs and target-specific uncertainty.
method Bayesian feature selection targeting specific outcomes.
result Feature selection for a specific target is superior to common methods.
Algorithms typically come with tunable parameters that have a considerable impact on the computational resources they consume. Too often, practitioners must hand-tune the parameters, a tedious and error-prone task. A recent line of research provides algorithms that return nearly-optimal parameters from within a finite …
A new method finds DAG models without ground truth.
problem Finding DAG models without ground truth.
method Markov Checker test and Cross-Algorithm Frugality Search (CAFS).
result Models can be chosen without ground truth.
Optimal transport aggregation combines distributed MoE models efficiently.
problem Combining local MoE models trained on distributed datasets.
method Optimal transport for minimizing divergence between local and global estimators, with MM algorithm for optimization.
result Aggregated estimator achieves performance comparable to centralized training but with reduced computation time.
Unimodal approach for group-level emotion recognition without individual features.
problem Privacy issues in individual-based emotion recognition models.
method Frugal approach using global features, state-of-the-art and synthetic corpora.
result 59.13% accuracy on VGAF test set, 11th place in EmotiW Challenge 2020.
ASPIRE improves amortized posterior inference for Bayesian inverse problems.
problem Bayesian inverse problems are computationally challenging due to uncertainty quantification.
method Iterative refinement of amortized posteriors using physics-based and summary statistics.
result ASPIRE achieves better posterior approximations with minimal extra computations.
When building large-scale machine learning (ML) programs, such as big topic models or deep neural nets, one usually assumes such tasks can only be attempted with industrial-sized clusters with thousands of nodes, which are out of reach for most practitioners or academic researchers. We consider this challenge in the co…