New methodology controls synthetic data bias for neural program synthesis.
problem Deep networks generalize poorly to certain data distributions when trained on synthetic examples.
method Proposes a new methodology to control and evaluate the bias of synthetic data distributions over programs and specifications.
result Training deep networks on controlled synthetic data distributions leads to improved cross-distribution generalization performance.
This article introduces the degenerate special Lagrangian equation (DSL) and develops the basic analytic tools to construct and study its solutions. The DSL governs geodesics in the space of positive graph Lagrangians in Cn. Existence of geodesics in the space of positive Lagrangians is an important step in…
DSL uses supervised learning to optimize portfolios, improving stability and performance.
problem Optimizing robust portfolios in financial markets.
method DSL reframes portfolio construction as a supervised learning problem, using cross-entropy loss and optimizing Sharpe or Sortino ratios. Deep Ensemble methods are employed to reduce variance.
result DSL outperforms traditional and machine learning methods, achieving higher median returns and more stable risk-adjusted performance.
New findings on convexity of special Lagrangian geodesics.
problem Convexity of special Lagrangian geodesics in space-time.
method Space-time coordinate transformation preserving Lagrangian angle, leading to C2 estimate. result Subsolutions in all branches of the degenerate special Lagrangian equation are bi-convex.
The process of designing neural architectures requires expert knowledge and extensive trial and error. While automated architecture search may simplify these requirements, the recurrent neural network (RNN) architectures generated by existing methods are limited in both flexibility and components. We propose a domain-s…
Proposes PA-DSL for correcting noisy human labels in automated data labeling.
problem Noisy human labels in automated data labeling.
method Uses adjudicated cases to correct noisy human labels and debias analyses.
result Maintains nominal coverage and reduces RMSE by 10-17% relative to using only adjudicated labels.
SED integrates synthesis, execution, and debugging for neural program synthesis.
problem Challenges in synthesizing complex programs that match specifications.
method SED combines synthesis, execution, and debugging to improve neural program generation.
result SED reduces error rates and outperforms standard decoding methods.
DSL estimates heterogeneous treatment effects over time in survival settings.
problem Complicated by right censoring and time-varying treatment effects.
method Deep survival learner (DSL) for estimating CATEs over a clinically relevant time spectrum.
result DSL reveals heterogeneity in perioperative chemotherapy effects over time.
We show that the degenerate special Lagrangian equation, recently introduced by Rubinstein-Solomon, induces a global equation on every Riemannian manifold, and that for certain associated geometries this equation governs, as it does in the Euclidean setting, geodesics in the space of positive Lagrangians. For example, …
DynamicPPL speeds up probabilistic modeling in Julia.
problem Developing and executing complex dynamic probabilistic models efficiently.
method Modular Julia library with a DSL, tracing data structure, and contextual dispatch.
result Achieves computational performance close to or better than Stan.
IReEn reveals functionality of black-box agents via iterative neural synthesis.
problem Revealing the functionality of a black-box agent without privileged information.
method Iterative refinement of candidate programs using neural program synthesis.
result The approach finds a functional equivalent program in 78% of cases, outperforming state-of-the-art.
PLANS synthesizes programs from noisy inputs using neural specs and filtering.
problem Synthesizing robust programs from noisy, raw inputs.
method Hybrid model combining neural extraction and rule-based synthesis with noise filtering.
result State-of-the-art performance in diverse environments with no ground-truth training.
The goal in network state prediction (NSP) is to classify the global state (label) associated with features embedded in a graph. This graph structure encoding feature relationships is the key distinctive aspect of NSP compared to classical supervised learning. NSP arises in various applications: gene expression samples…
The results of data mining endeavors are majorly driven by data quality. Throughout these deployments, serious show-stopper problems are still unresolved, such as: data collection ambiguities, data imbalance, hidden biases in data, the lack of domain information, and data incompleteness. This paper is based on the prem…
New method uses imperfect LLM annotations for valid statistical inference in social science.
problem Inaccurate large language model annotations in social science research.
method Design-based supervised learning (DSL) combining imperfect LLM surrogates with gold-standard labels.
result DSL provides valid statistical inference with comparable predictive accuracy to existing methods.
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…
This paper characterizes and designs loss functions for robust classification with abstention.
problem Ensuring robustness against adversarial attacks and knowing when to abstain from prediction.
method Proposes adversarial robust reject option loss and characterizes surrogates for calibration.
result Shifted Double Ramp Loss and Shifted Double Sigmoid Loss satisfy the calibration conditions.
Recent work has shown how to embed differentiable optimization problems (that is, problems whose solutions can be backpropagated through) as layers within deep learning architectures. This method provides a useful inductive bias for certain problems, but existing software for differentiable optimization layers is rigid…
Study optimal rates for multiclass classification, resolving open questions.
problem Optimal rates for multiclass classification with any label space.
method Establishes optimal rates for all hypothesis classes, defining new tree structures.
result Optimal rates for multiclass classification with no infinite DSL trees.
A new approach for specifying and synthesizing subroutines for optimizing metrics.
problem Specifying and optimizing subroutines for various metrics.
method Formalizing programming by rewards (PBR), using continuous-optimization techniques to synthesize decision functions as if-then-else programs.
result Synthesized decision functions are optimal in cases when rewards have nice properties.
LLM agents discover cryptocurrency factors under reproducible constraints.
problem Flexibility of LLM agents in empirical discovery leads to uncontrolled search.
method Sequential hypothesis search with fixed data splits and portfolio tests.
result Ridge-combined portfolio achieves 44.55% annualized return in out-of-sample period.
Variational inference improves hierarchical imitation learning of control programs.
problem Learning structured control policies from demonstrations.
method Variational inference for discovering hierarchical structure in observation-action traces.
result Variational inference leads to more efficient and generalized control policies.
We introduce SPFlow, an open-source Python library providing a simple interface to inference, learning and manipulation routines for deep and tractable probabilistic models called Sum-Product Networks (SPNs). The library allows one to quickly create SPNs both from data and through a domain specific language (DSL). It e…
Improved genetic programming by optimizing mutation operators for continuous program search.
problem Small syntactic mutations in genetic programming can lead to unpredictable behavioral shifts.
method Learned a compact trading-strategy DSL, created a block-factorized embedding, and designed geometry-compiled mutation operators.
result Geometry-compiled mutation operators discover strong strategies using fewer evaluations and achieve higher Sharpe ratios.