arXiv subjects

Thiago Serra

Publications and source records attributed to Thiago Serra.

At least 19 recordsLinked to original sources

arXiv2026

Optimal Combinatorial Testing with Constraints: The Balancing Act

Imagine that you are in front of a cockpit with several on-off buttons. If you were to thoroughly test it, you would need to try a prohibitive number of configurations. But since most bugs in practice can be isolated to interactions among few components, having tests that cover every possible pairwise configuration is a good start. However, this is a problem that goes from easy to NP-hard as soon as some pairwise configurations are forbidden. In this paper, we revisit unconstrained combinatorial testing with pairwise coverage on binary parameters and contrast it with the constrained case, showing and conjecturing properties that either are upheld or change from one to the other. In particular, we discuss the extent to which it remains a good idea -- and sometimes indeed optimal -- to have every button almost as many times on as off to minimize testing. We propose the first exact algorithm based on integer programming and a faster heuristic that often produces optimal solutions, both outperforming or competitive with their baselines.

math.OC↗

arXiv2025

Optimization over Trained Neural Networks: Going Large with Gradient-Based Algorithms

When optimizing a nonlinear objective, one can employ a neural network as a surrogate for the nonlinear function. However, the resulting optimization model can be time-consuming to solve globally with exact methods. As a result, local search that exploits the neural-network structure has been employed to find good solutions within a reasonable time limit. For such methods, a lower per-iteration cost is advantageous when solving larger models. The contribution of this paper is two-fold. First, we propose a gradient-based algorithm with lower per-iteration cost than existing methods. Second, we further adapt this algorithm to exploit the piecewise-linear structure of neural networks that use Rectified Linear Units (ReLUs). In line with prior research, our methods become competitive with -- and then dominant over -- other local search methods as the optimization models become larger.

math.OC↗

arXiv2025

Final Exam Scheduling at Bucknell University: A Case Study and Open-Source Tool

Problem Definition: Final exam scheduling is a common but challenging optimization problem. At Bucknell University, a small liberal arts institution, the problem is particularly complex and has historically required the Registrar's Office to spend months manually designing an exam schedule each semester. Methodology: We worked in close collaboration with the Registrar's Office. First, we created visualization tools to help their manual scheduling process. Then we designed integer programming models and heuristics to produce a portfolio of possible exam schedules. Finally, we developed open-source, user-friendly software, enabling their office to directly produce and adjust these schedules. Results and Managerial Implications: Our tools - both software and algorithms - are now in use at Bucknell University. This collaboration has led to both substantial time savings and improved schedules. Since the implementation of this project, for example, the proportion of students who have a back-to-back exam in a given semester has decreased from roughly a third to about 10%. Our tools are fully open-source and rely on an open-source optimization solver, and our approach garnered national media attention with an article in Inside Higher Ed.

math.OC↗

arXiv2025

What is the Best Way to Do Something? A Discreet Tour of Discrete Optimization

In mathematical optimization, we want to find the best possible solution for a decision-making problem. Curiously, these problems are harder to solve if they have discrete decisions. Imagine that you would like to buy chocolate: you can buy no chocolate or one chocolate bar, but typically you cannot buy just half of a bar. Now imagine that you could also buy many other items, and that you need to meet nutritional needs while minimizing the grocery bill. With more options and more demands, finding the best solution becomes trickier. But since many real-world settings benefit from mathematical optimization, such as scheduling trains and flights, planning truck deliveries, and making better investment decisions, these problems are widely studied in a branch of mathematics called Operations Research (OR). Sometimes we can simply write the mathematical model and find an optimal solution with OR software, but for larger problems we may need to develop new mathematical models and even write our own algorithms. We explore both cases with a simple and well-known problem (the traveling salesperson problem), some computer programming (in Python), and software that is free for academic use (Gurobi). All the code and data used is available at: https://github.com/thserra/discreet

math.OC↗

arXiv2025

Making a Case for Research Collaboration Between Artificial Intelligence and Operations Research Experts

In 2021, INFORMS, ACM SIGAI, and the Computing Community Consortium (CCC) hosted three workshops to explore synergies between Artificial Intelligence (AI) and Operations Research (OR) to improve decision-making. The workshops aimed to create a unified research vision for AI/OR collaboration, focusing on overcoming cultural differences and maximizing societal impact. The first two workshops addressed technological innovations, applications, and trustworthy AI development, while the final workshop highlighted specific areas for AI/OR integration. Participants discussed "Challenge Problems" and strategies for combining AI and OR techniques. This report outlines five key recommendations to enhance AI/OR collaboration: 1) Funding Opportunities, 2) Joint Education, 3) Long-term Research Programs, 4) Aligning Conferences/Journals, and 5) Benchmark Creation.

cs.CY↗

arXiv2025

Optimization over Trained (and Sparse) Neural Networks: A Surrogate within a Surrogate

In constraint learning, we use a neural network as a surrogate for part of the constraints or of the objective function of an optimization model. However, the tractability of the resulting model is heavily influenced by the size of the neural network used as a surrogate. One way to obtain a more tractable surrogate is by pruning the neural network first. In this work, we consider how to approach the setting in which the neural network is actually a given: how can we solve an optimization model embedding a large and predetermined neural network? We propose surrogating the neural network itself by pruning it, which leads to a sparse and more tractable optimization model, for which we hope to still obtain good solutions with respect to the original neural network. For network verification and function maximization models, that indeed leads to better solutions within a time limit, especially -- and surprisingly -- if we skip the standard retraining step known as finetuning. Hence, a pruned network with worse inference for lack of finetuning can be a better surrogate.

math.OC↗

arXiv2024

Computational Tradeoffs of Optimization-Based Bound Tightening in ReLU Networks

The use of Mixed-Integer Linear Programming (MILP) models to represent neural networks with Rectified Linear Unit (ReLU) activations has become increasingly widespread in the last decade. This has enabled the use of MILP technology to test-or stress-their behavior, to adversarially improve their training, and to embed them in optimization models leveraging their predictive power. Many of these MILP models rely on activation bounds. That is, bounds on the input values of each neuron. In this work, we explore the tradeoff between the tightness of these bounds and the computational effort of solving the resulting MILP models. We provide guidelines for implementing these models based on the impact of network structure, regularization, and rounding.

math.OC↗

arXiv2024

Optimization Over Trained Neural Networks: Taking a Relaxing Walk

Besides training, mathematical optimization is also used in deep learning to model and solve formulations over trained neural networks for purposes such as verification, compression, and optimization with learned constraints. However, solving these formulations soon becomes difficult as the network size grows due to the weak linear relaxation and dense constraint matrix. We have seen improvements in recent years with cutting plane algorithms, reformulations, and an heuristic based on Mixed-Integer Linear Programming (MILP). In this work, we propose a more scalable heuristic based on exploring global and local linear relaxations of the neural network model. Our heuristic is competitive with a state-of-the-art MILP solver and the prior heuristic while producing better solutions with increases in input, depth, and number of neurons.

math.OC↗

arXiv2023

When Deep Learning Meets Polyhedral Theory: A Survey

In the past decade, deep learning became the prevalent methodology for predictive modeling thanks to the remarkable accuracy of deep neural networks in tasks such as computer vision and natural language processing. Meanwhile, the structure of neural networks converged back to simpler representations based on piecewise constant and piecewise linear functions such as the Rectified Linear Unit (ReLU), which became the most commonly used type of activation function in neural networks. That made certain types of network structure $\unicode{x2014}$such as the typical fully-connected feedforward neural network$\unicode{x2014}$ amenable to analysis through polyhedral theory and to the application of methodologies such as Linear Programming (LP) and Mixed-Integer Linear Programming (MILP) for a variety of purposes. In this paper, we survey the main topics emerging from this fast-paced area of work, which bring a fresh perspective to understanding neural networks in more detail as well as to applying linear optimization techniques to train, verify, and reduce the size of such networks.

math.OC↗

arXiv2023

Getting Away with More Network Pruning: From Sparsity to Geometry and Linear Regions

One surprising trait of neural networks is the extent to which their connections can be pruned with little to no effect on accuracy. But when we cross a critical level of parameter sparsity, pruning any further leads to a sudden drop in accuracy. This drop plausibly reflects a loss in model complexity, which we aim to avoid. In this work, we explore how sparsity also affects the geometry of the linear regions defined by a neural network, and consequently reduces the expected maximum number of linear regions based on the architecture. We observe that pruning affects accuracy similarly to how sparsity affects the number of linear regions and our proposed bound for the maximum number. Conversely, we find out that selecting the sparsity across layers to maximize our bound very often improves accuracy in comparison to pruning as much with the same sparsity in all layers, thereby providing us guidance on where to prune.

cs.LG↗

arXiv2022

Recall Distortion in Neural Network Pruning and the Undecayed Pruning Algorithm

Pruning techniques have been successfully used in neural networks to trade accuracy for sparsity. However, the impact of network pruning is not uniform: prior work has shown that the recall for underrepresented classes in a dataset may be more negatively affected. In this work, we study such relative distortions in recall by hypothesizing an intensification effect that is inherent to the model. Namely, that pruning makes recall relatively worse for a class with recall below accuracy and, conversely, that it makes recall relatively better for a class with recall above accuracy. In addition, we propose a new pruning algorithm aimed at attenuating such effect. Through statistical analysis, we have observed that intensification is less severe with our algorithm but nevertheless more pronounced with relatively more difficult tasks, less complex models, and higher pruning ratios. More surprisingly, we conversely observe a de-intensification effect with lower pruning ratios, which indicates that moderate pruning may have a corrective effect to such distortions.

cs.LG↗

arXiv2022

The Combinatorial Brain Surgeon: Pruning Weights That Cancel One Another in Neural Networks

Neural networks tend to achieve better accuracy with training if they are larger -- even if the resulting models are overparameterized. Nevertheless, carefully removing such excess parameters before, during, or after training may also produce models with similar or even improved accuracy. In many cases, that can be curiously achieved by heuristics as simple as removing a percentage of the weights with the smallest absolute value -- even though magnitude is not a perfect proxy for weight relevance. With the premise that obtaining significantly better performance from pruning depends on accounting for the combined effect of removing multiple weights, we revisit one of the classic approaches for impact-based pruning: the Optimal Brain Surgeon(OBS). We propose a tractable heuristic for solving the combinatorial extension of OBS, in which we select weights for simultaneous removal, as well as a systematic update of the remaining weights. Our selection method outperforms other methods under high sparsity, and the weight update is advantageous even when combined with the other methods.

cs.LG↗

arXiv2022

Training Thinner and Deeper Neural Networks: Jumpstart Regularization

Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization of the loss function to make the effect of most parameters negligible. In this work, we propose instead to use regularization for preventing neurons from dying or becoming linear, a technique which we denote as jumpstart regularization. In comparison to conventional training, we obtain neural networks that are thinner, deeper, and - most importantly - more parameter-efficient.

cs.LG↗

arXiv2022

Optimal Decision Diagrams for Classification

Decision diagrams for classification have some notable advantages over decision trees, as their internal connections can be determined at training time and their width is not bound to grow exponentially with their depth. Accordingly, decision diagrams are usually less prone to data fragmentation in internal nodes. However, the inherent complexity of training these classifiers acted as a long-standing barrier to their widespread adoption. In this context, we study the training of optimal decision diagrams (ODDs) from a mathematical programming perspective. We introduce a novel mixed-integer linear programming model for training and demonstrate its applicability for many datasets of practical importance. Further, we show how this model can be easily extended for fairness, parsimony, and stability notions. We present numerical analyses showing that our model allows training ODDs in short computational times, and that ODDs achieve better accuracy than optimal decision trees, while allowing for improved stability without significant accuracy losses.

cs.LG↗

arXiv2022

Seamless Multimodal Transportation Scheduling

Ride-hailing services have expanded the role of shared mobility in passenger transportation systems, creating new markets and creative planning solutions for major urban centers. In this paper, we consider their use for the first-mile or last-mile passenger transportation in coordination with a mass transit service to provide a seamless multimodal transportation experience for the user. A system that provides passengers with predictable information on travel and waiting times in their commutes is immensely valuable. We envision that the passengers will inform the system of their desired travel and arrival windows so that the system can jointly optimize the schedules of passengers. The problem we study balances minimizing travel time and the number of trips taken by the last-mile vehicles, so that long-term planning, maintenance, and environmental impact are all taken into account. We focus on the case where the last-mile service aggregates passengers by destination. We show that this problem is NP-hard, and propose a decision diagram-based branch-and-price decomposition model that can solve instances of real-world size (10,000 passengers spread over an hour, 50 last-mile destinations, 600 last-mile vehicles) in computational time (~1 minute) that is orders-of-magnitude faster than other methods appearing in the literature. Our experiments also indicate that aggregating passengers by destination on the last-mile service provides high-quality solutions to more general settings.

math.OC↗

arXiv2021

Scaling Up Exact Neural Network Compression by ReLU Stability

We can compress a rectifier network while exactly preserving its underlying functionality with respect to a given input domain if some of its neurons are stable. However, current approaches to determine the stability of neurons with Rectified Linear Unit (ReLU) activations require solving or finding a good approximation to multiple discrete optimization problems. In this work, we introduce an algorithm based on solving a single optimization problem to identify all stable neurons. Our approach is on median 183 times faster than the state-of-art method on CIFAR-10, which allows us to explore exact compression on deeper (5 x 100) and wider (2 x 800) networks within minutes. For classifiers trained under an amount of L1 regularization that does not worsen accuracy, we can remove up to 56% of the connections on the CIFAR-10 dataset. The code is available at the following link, https://github.com/yuxwind/ExactCompression.

cs.LG↗

arXiv2021

Template-based Minor Embedding for Adiabatic Quantum Optimization

Quantum Annealing (QA) can be used to quickly obtain near-optimal solutions for Quadratic Unconstrained Binary Optimization (QUBO) problems. In QA hardware, each decision variable of a QUBO should be mapped to one or more adjacent qubits in such a way that pairs of variables defining a quadratic term in the objective function are mapped to some pair of adjacent qubits. However, qubits have limited connectivity in existing QA hardware. This has spurred work on preprocessing algorithms for embedding the graph representing problem variables with quadratic terms into the hardware graph representing qubits adjacencies, such as the Chimera graph in hardware produced by D-Wave Systems. In this paper, we use integer linear programming to search for an embedding of the problem graph into certain classes of minors of the Chimera graph, which we call template embeddings. One of these classes corresponds to complete bipartite graphs, for which we show the limitation of the existing approach based on minimum Odd Cycle Transversals (OCTs). One of the formulations presented is exact, and thus can be used to certify the absence of a minor embedding using that template. On an extensive test set consisting of random graphs from five different classes of varying size and sparsity, we can embed more graphs than a state-of-the-art OCT-based approach, our approach scales better with the hardware size, and the runtime is generally orders of magnitude smaller.

cs.DS↗

arXiv2020

Lossless Compression of Deep Neural Networks

Deep neural networks have been successful in many predictive modeling tasks, such as image and language recognition, where large neural networks are often used to obtain good accuracy. Consequently, it is challenging to deploy these networks under limited computational resources, such as in mobile devices. In this work, we introduce an algorithm that removes units and layers of a neural network while not changing the output that is produced, which thus implies a lossless compression. This algorithm, which we denote as LEO (Lossless Expressiveness Optimization), relies on Mixed-Integer Linear Programming (MILP) to identify Rectified Linear Units (ReLUs) with linear behavior over the input domain. By using L1 regularization to induce such behavior, we can benefit from training over a larger architecture than we would later use in the environment where the trained neural network is deployed.

cs.LG↗