DeepEvolution improves testing of deep neural networks by generating diverse test cases.
problem Lack of diversity in generated test cases from random fuzzing or transformations.
method Search-based approach using metaheuristics to ensure diversity in test cases.
result DeepEvolution significantly increases neuronal coverage and detects latent defects.
Neural circuit model re-purposed for robotic control tasks.
problem Learning simple robotic control tasks.
method Re-purposing a biological neural circuit model to control robotic tasks using a search-based optimization algorithm.
result Neuronal Circuit Policies (NCPs) perform on par and in some cases surpass contemporary deep learning models with fewer parameters and interpretable dynamics.
PackIt creates a virtual space for testing geometric planning skills.
problem Evaluating geometric planning abilities in virtual environments.
method Developed a virtual environment, PackIt, for geometric planning tasks.
result Demonstrated the effectiveness of various methods for geometric planning.
This paper presents preliminary work on learning the search heuristic for the optimal motion planning for automated driving in urban traffic. Previous work considered search-based optimal motion planning framework (SBOMP) that utilized numerical or model-based heuristics that did not consider dynamic obstacles. Optimal…
SPoC uses search to translate pseudocode into correct programs with error localization.
problem Mapping pseudocode to functionally correct long programs.
method Search-based approach guided by compilation errors for credit assignment.
result Search improves synthesis success rate from 25.6% to 44.7%.
Neural A* uses machine learning to improve path planning efficiency.
problem Challenges in applying machine learning to search-based path planning.
method Reformulated A* search as a differentiable network coupled with a convolutional encoder.
result Neural A* outperforms state-of-the-art planners in optimality and efficiency.
Neural model with parameterized algorithms improves graph CO problem solving.
problem Solving NP-hard graph combinatorial optimization problems efficiently and accurately.
method Combining neural models and parameterized algorithms to identify and handle hard and easy parts of CO instances.
result Framework produces superior solution quality and out-of-distribution generalization.
Neurally-Guided Structure Inference combines search and data-driven methods for efficient, robust structure inference.
problem Combining the advantages of exhaustive search and data-driven methods for structure inference.
method Neurally-Guided Structure Inference (NG-SI) uses a neural network to guide hierarchical search over structures.
result NG-SI outperforms search-based and data-driven methods on probabilistic matrix decomposition and symbolic program parsing.
Addressing the issue of SVMs parameters optimization, this study proposes an efficient memetic algorithm based on Particle Swarm Optimization algorithm (PSO) and Pattern Search (PS). In the proposed memetic algorithm, PSO is responsible for exploration of the search space and the detection of the potential regions with…
XNAS optimizes neural architecture search using expert advice theory.
problem Optimizing neural architecture selection with minimal regret.
method Uses prediction with expert advice theory and dynamic architecture wiping.
result Achieves optimal worst-case regret and state-of-the-art results.
Efficiently reduces computational burden of rollout acquisition functions in Bayesian optimization.
problem Expensive computation of rollout acquisition functions in Bayesian optimization.
method Combines quasi-Monte Carlo, common random numbers, and control variates to reduce computational burden. Formulates a policy-search approach to eliminate the need to optimize the rollout acquisition function.
result Significant reduction in computational burden of rollout acquisition functions.
New sparse GP model learns compositional kernels efficiently.
problem Learning accurate Gaussian Process models with complex kernel structures.
method MultiSVGP model with Horseshoe prior for kernel selection.
result Our model provides better fit and faster computation for large-scale data.
BPI models 2D patterns on multiple planes and 3D scene from a single image.
problem Understanding and editing images with multiple 2D planes and 3D scene from a single image.
method Box Program Induction (BPI) with neural networks and search-based algorithm.
result Holistic, structured scene representation enables 3D-aware image editing.
NeuRewriter learns to choose and rewrite heuristics in combinatorial problems.
problem Time-consuming tuning of heuristics in combinatorial optimization.
method NeuRewriter uses reinforcement learning to learn a policy for picking heuristics and rewriting solutions.
result NeuRewriter outperforms existing methods in various combinatorial tasks.
Statistical relational frameworks such as Markov logic networks and probabilistic soft logic (PSL) encode model structure with weighted first-order logical clauses. Learning these clauses from data is referred to as structure learning. Structure learning alleviates the manual cost of specifying models. However, this be…
SIM models user interests from long sequential behavior data, improving click-through rate prediction.
problem Challenges in capturing user interests with long user behavior sequences.
method SIM uses a cascaded search paradigm with two units: General Search Unit and Exact Search Unit.
result SIM achieves significant CTR and RPM lifts in Alibaba's display advertising system.
A new Tabu Search method optimizes neural network architecture.
problem Optimizing the architecture of deep neural networks.
method Combines Tabu Search and GDM for automatic architecture selection.
result Demonstrates Tabu Search can automatically select optimal neural network architecture.
A new algorithm optimizes complex systems by intelligently using different levels of information.
problem Optimizing systems with multiple, cost-dependent information sources.
method Proposes MF-MI-Greedy, a principled algorithm using additive Gaussian processes and cost-sensitive mutual information gain.
result MF-MI-Greedy achieves low regret and demonstrates strong empirical performance.
Dimension reduction and variable selection are performed routinely in case-control studies, but the literature on the theoretical aspects of the resulting estimates is scarce. We bring our contribution to this literature by studying estimators obtained via L1 penalized likelihood optimization. We show that the optimize…
Yau's Affine Normal Descent optimizes smooth unconstrained problems with geometrically adapted directions.
problem Optimizing smooth unconstrained problems with geometrically adapted directions.
method Yau's Affine Normal Descent (YAND) uses the equi-affine normal of level-set hypersurfaces as search directions.
result YAND converges globally under standard smoothness assumptions and locally quadratically near nondegenerate minimizers.
The classification of MRI images according to the anatomical field of view is a necessary task to solve when faced with the increasing quantity of medical images. In parallel, advances in deep learning makes it a suitable tool for computer vision problems. Using a common architecture (such as AlexNet) provides quite go…
Two approaches scale up DNN optimization for diverse edge devices.
problem Optimizing DNNs for edge devices with varying performance requirements.
method Reuse performance predictors on proxy devices and build scalable predictors.
result Optimized DNN designs for many different edge devices without lengthy optimization.
In many sequential decision making tasks, it is challenging to design reward functions that help an RL agent efficiently learn behavior that is considered good by the agent designer. A number of different formulations of the reward-design problem, or close variants thereof, have been proposed in the literature. In this…
BOOOM optimizes orthonormal matrices without needing gradients.
problem Optimizing over the Stiefel manifold in non-convex, non-smooth settings.
method Global Givens rotation-based parametrization and Recursive Modified Pattern Search.
result BOOOM achieves strong performance across various optimization problems.
Approximate inference in high-dimensional, discrete probabilistic models is a central problem in computational statistics and machine learning. This paper describes discrete particle variational inference (DPVI), a new approach that combines key strengths of Monte Carlo, variational and search-based techniques. DPVI is…
A framework for efficient multi-objective optimization using entropy search.
problem Optimizing expensive black-box functions with multiple objectives.
method Output space entropy search (OSE) to minimize resource cost.
result Improves efficiency and accuracy in multi-objective optimization.
New method solves stochastic optimization problems with random models.
problem Optimizing stochastic objectives with deterministic constraints.
method Trust-Region Sequential Quadratic Programming with random model.
result Global convergence guarantees for first- and second-order stationary points.
AutoAlpha efficiently discovers effective alpha factors for quantitative investment.
problem Mining effective alpha factors for successful quantitative investment models.
method Hierarchical evolutionary algorithm with PCA-QD search, warm start, and replacement methods.
result AutoAlpha discovers and generates effective formulaic alphas for portfolio optimization.
Paper tackles multi-action policy learning from observational data.
problem Learning policies from observational data with multiple actions.
method Semi-parametric inference theory and algorithmic approaches.
result Achieves asymptotically minimax-optimal regret for multi-action policy learning.
BASE framework learns task-agnostic architectures for faster NAS.
problem High computational cost in Neural Architecture Search.
method Bayesian Meta Architecture Search (BASE) framework.
result Found models achieving 25.7% top-1 error and 8.1% top-5 error in less than an hour.
Accurate real-time tracking of influenza outbreaks helps public health officials make timely and meaningful decisions that could save lives. We propose an influenza tracking model, ARGO (AutoRegression with GOogle search data), that uses publicly available online search data. In addition to having a rigorous statistica…
RSO uses random weight perturbations to train deep networks without gradients.
problem Training deep neural networks efficiently and without gradient information.
method RSO is a gradient-free Markov Chain Monte Carlo approach that updates weights based on mini-batch loss reduction.
result RSO achieves high accuracy (99.1% on MNIST) with significantly fewer updates than traditional methods.
Algorithm provides fair clustering guarantees for k-means and k-median.
problem Ensuring fair clustering guarantees for every point in a dataset.
method Local search algorithm for k-median and k-means, with individual fairness as a key metric. result Achieves fair clustering with a constant factor approximation of the optimal solution.
Paper trains models to resist string transformations.
problem Vulnerability of NLP models to adversarial string transformations.
method Combines search and abstraction techniques for robust training.
result Trained models resist combinations of user-defined transformations.
UNAS combines DNAS and RL for efficient architecture search.
problem Discovering high accuracy or low latency neural architectures.
method Unified framework combining differentiable and reinforcement learning approaches.
result UNAS achieves state-of-the-art accuracy on CIFAR-10, CIFAR-100, and ImageNet datasets.
INT benchmark tests theorem proving agents' ability to generalize to unseen theorems.
problem Evaluating theorem proving agents' ability to generalize to unseen theorems.
method INT benchmark based on a theorem generation and proof procedure with adjustable knobs for measuring 6 types of generalization.
result MCTS can help agents prove new theorems.
ISMCTS-BR learns best responses in large games, approximating worst-case performance.
problem Learning robustness to worst-case outcomes in large games.
method ISMCTS-BR, a scalable search-based algorithm for deep reinforcement learning.
result ISMCTS-BR approximates worst-case performance in large games.
This paper improves neural architecture search by focusing on novelty-driven sampling.
problem Lack of positive correlation between validation accuracy and test accuracy in One-Shot NAS.
method Single-path supernet with optimized weights and novelty search for architecture sampling.
result Novelty search method achieves state-of-the-art test error rate on CIFAR-10.
P3I learns holistic scene representations from a single image.
problem Inferring camera poses, object locations, and global scene structures from a single image.
method Combines search-based and gradient-based algorithms.
result P3I outperforms baselines on various image manipulation tasks.
Deep neural network estimates support of sparse signals for improved phase retrieval.
problem Sparse phase retrieval from Fourier magnitudes with support estimation.
method Trained deep neural network (DNN) provides extended support estimate E larger than the support T. result DNN-based support estimation improves signal reconstruction performance with lower complexity.
Wireless systems perform rate adaptation to transmit at highest possible instantaneous rates. Rate adaptation has been increasingly granular over generations of wireless systems. The base-station uses SINR and packet decode feedback called acknowledgement/no acknowledgement (ACK/NACK) to perform rate adaptation. SINR i…
Dynamic cell-free networks reduce complexity in serving many devices with distributed APs and DRL.
problem Designing efficient cell-free networks with many devices and APs.
method Dynamic architecture, SIC, DAS, DRL for optimization.
result DRL significantly improves performance in dynamic cell-free networks.
Enzyme sequences and structures are routinely used in the biological sciences as queries to search for functionally related enzymes in online databases. To this end, one usually departs from some notion of similarity, comparing two enzymes by looking for correspondences in their sequences, structures or surfaces. For a…
Paper proposes a new model and methods for robustly de-interleaving HMP mixtures.
problem Lack of robustness to non-ideal situations in existing HMP mixtures de-interleaving methods.
method Designs a generative model, formulates de-interleaving as posterior inference, develops exact and approximate inference methods, derives error probability bounds.
result Proposed methods are highly effective and robust for non-ideal situations, outperforming baseline methods.
New test assesses reliability of auto-generated features.
problem Lack of reliable assessment for auto-generated features.
method Selective inference framework for statistical testing.
result Proposes a statistical test for auto-generated features in linear models.
Paper presents a self-supervised method to infer road lane networks.
problem Difficult and costly to create lane maps for autonomous vehicles.
method Self-supervised learning using neural and search-based model.
result Model can generalize to new road layouts, unlike previous approaches.
Transformers learn to predict chess moves with surprising accuracy and strength.
problem Training transformers on chess to predict moves accurately.
method Large-scale chess dataset (10M games), supervised learning with up to 270M parameters.
result Transformers can predict action-values for novel boards with high accuracy.
DAEGEN generates adversarial inputs for neural networks using a black-box differential technique.
problem Generating adversarial inputs that highlight differences between neural network models.
method DAEGEN uses a local search-based optimization algorithm to find difference-inducing adversarial examples (DIAEs).
result DAEGEN is the first black-box differential technique for adversarial input generation and performs well compared to existing methods.