We consider the problem of learning multiple tasks in a continual learning setting in which data from different tasks is presented to the learner in a streaming fashion. A key challenge in this... Show moreWe consider the problem of learning multiple tasks in a continual learning setting in which data from different tasks is presented to the learner in a streaming fashion. A key challenge in this setting is the so-called "catastrophic forgetting problem", in which the performance of the learner in an "old task" decreases when subsequently trained on a "new task". Existing continual learning methods, such as Averaged Gradient Episodic Memory (A-GEM) and Orthogonal Gradient Descent (OGD), address catastrophic forgetting by minimizing the loss for the current task without increasing the loss for previous tasks. However, these methods assume the learner knows when the task changes, which is unrealistic in practice. In this paper, we alleviate the need to provide the algorithm with information about task changes by using an online clustering-based approach on a dynamically updated finite pool of samples or gradients. We thereby successfully counteract catastrophic forgetting in one of the hardest settings, namely: domain-incremental learning, a setting for which the problem was previously unsolved. We showcase the benefits of our approach by applying these ideas to projection-based methods, such as A-GEM and OGD, which lead to task-agnostic versions of them. Experiments on real datasets demonstrate the effectiveness of the proposed strategy and its promising performance compared to state-of-the-art methods. Show less
Stein, N. van; Long, F.; Frenzel, M.; Krause, P.; Gitterle, M.; Bäck, T.H.W. 2023
Decisions made in the early phases of ship design have a large influence on the capital and operational expenses of a vessel. In order to support decision making in this phase, big data and machine... Show moreDecisions made in the early phases of ship design have a large influence on the capital and operational expenses of a vessel. In order to support decision making in this phase, big data and machine learning techniques can be of great use. This work shows how Explainable Artificial Intelligence (XAI) and Global Sensitivity Analysis (GSA) combined with Autonomous Identification System (AIS) and static ship data can be used to find important design characteristics of ships. A data collection framework is setup that collects AIS data over a five month time period. Static ship design data is used to predict performance related target features that are calculated from AIS data. By applying XAI and GSA methods to the regression models that predict these target features, insight can be gained on how design features influence the performance of ships. Experiments showed that for most ship types, the overall length is the most important design feature for speed related target features. Besides the overall length, the draught also has a significant impact on the rotation capabilities. Show less
Huang, Q.; Winter, R. de; Stein, N. van; Bäck, T.H.W.; Kononova, A.V. 2022
Decades of progress in simulation-based surrogate-assisted optimization and unprecedented growth in computational power have enabled researchers and practitioners to optimize previously intractable... Show moreDecades of progress in simulation-based surrogate-assisted optimization and unprecedented growth in computational power have enabled researchers and practitioners to optimize previously intractable complex engineering problems. This paper investigates the possible benefit of a concurrent utilization of multiple simulation-based surrogate models to solve complex discrete optimization problems. To fulfill this, the so-called Self-Adaptive Multi-surrogate Assisted Efficient Global Optimization algorithm (SAMA-DiEGO), which features a two-stage online model management strategy, is proposed and further benchmarked on fifteen binary-encoded combinatorial and fifteen ordinal problems against several state-of-the-art non-surrogate or single surrogate assisted optimization algorithms. Our findings indicate that SAMA-DiEGO can rapidly converge to better solutions on a majority of the test problems which shows the feasibility and advantage of using multiple surrogate models in optimizing discrete problems Show less
Neural Architecture Search (NAS) aims to optimize deep neural networks' architecture for better accuracy or smaller computational cost and has recently gained more research interests. Despite... Show moreNeural Architecture Search (NAS) aims to optimize deep neural networks' architecture for better accuracy or smaller computational cost and has recently gained more research interests. Despite various successful approaches proposed to solve the NAS task, the landscape of it, along with its properties, are rarely investigated. In this paper, we argue for the necessity of studying the landscape property thereof and propose to use the so-called Exploratory Landscape Analysis (ELA) techniques for this goal. Taking a broad set of designs of the deep convolutional network, we conduct extensive experimentation to obtain their performance. Based on our analysis of the experimental results, we observed high similarities between well-performing architecture designs, which is then used to significantly narrow the search space to improve the efficiency of any NAS algorithm. Moreover, we extract the ELA features over the NAS landscapes on three common image classification data sets, MNIST, Fashion, and CIFAR-10, which shows that the NAS landscape can be distinguished for those three data sets. Also, when comparing to the ELA features of the well-known Black-Box optimization Benchmarking (BBOB) problem set, we found out that the NAS landscapes surprisingly form a new problem class on its own, which can be separated from all 24 BBOB problems. Given this interesting observation, we, therefore, state the importance of further investigation on selecting an efficient optimizer for the NAS landscape as well as the necessity of augmenting the current benchmark problem set. Show less
Wang, Y.; Stein, N. van; Bäck, T.H.W.; Emmerich, M.T.M. 2020
A customized multi-objective evolutionary algorithm (MOEA) is proposed for the flexible job shop scheduling problem (FJSP) with three objectives: makespan, total workload, critical workload. In... Show moreA customized multi-objective evolutionary algorithm (MOEA) is proposed for the flexible job shop scheduling problem (FJSP) with three objectives: makespan, total workload, critical workload. In general, the algorithm can be integrated with any standard MOEA. In this paper, it has been combined with NSGA-III to solve the state-of-the-art benchmark FJSPs, whereas an off-the-shelf implementation of NSGA-III is not capable of solving them. Most importantly, we use the various algorithm adaptations to enhance the performance of our algorithm. To be specific, it uses smart initialization approaches to enrich the first-generation population, and proposes new crossover operator to create a better diversity on the Pareto front approximation. The MIP-EGO configurator is adopted to automatically tune the mutation probabilities, which are important hyper-parameters of the algorithm. Furthermore, different local search strategies are employed to explore the neighborhood for better solutions. The experimental results from the combination of these techniques show the good performance as compared to classical evolutionary scheduling algorithms and it requires less computing budget. Even some previously unknown non-dominated solutions for the BRdata benchmark problems could be discovered. Show less
Wang, Y.; Stein, N. van; Bäck, T.H.W.; Emmerich, M.T.M. 2020
This paper provides a short summary of a novel algorithm tailored towards multi-objective flexible job shop scheduling problems (FJSP). The result shows that for challenging real-world problems in... Show moreThis paper provides a short summary of a novel algorithm tailored towards multi-objective flexible job shop scheduling problems (FJSP). The result shows that for challenging real-world problems in combinatorial optimization, off-the-shelf implementations of multi-objective optimization evolutionary algorithms (MOEAs) might not work, but by using various adaptations, these methods can be tailored to provide excellent results. This is demonstrated for a state of the art MOEA, that is NSGA-III, and the following adaptations: (1) initialization approaches to enrich the first-generation population, (2) various crossover operators to create a better diversity of offspring, (3) parameter tuning, to determine the optimal mutation probabilities, using the MIP-EGO configurator, (4) local search strategies to explore the neighborhood for better solutions. Using these measures, NSGA-III has been enabled to solve benchmark multi-objective FJSPs and experimental results show excellent performance. Show less