Neural architecture search (NAS) aims to discover network architectures with desired properties such as high accuracy or low latency. Recently, differentiable NAS (DNAS) has demonstrated promising results while maintaining a search cost orders of magnitude lower than reinforcement learning (RL) based NAS. However, DNAS models can only optimize differentiable loss functions in search, and they require an accurate differentiable approximation of non-differentiable criteria. In this work, we present UNAS, a unified framework for NAS, that encapsulates recent DNAS and RL-based approaches under one framework. Our framework brings the best of both worlds, and it enables us to search for architectures with both differentiable and non-differentiable criteria in one unified framework while maintaining a low search cost. Further, we introduce a new objective function for search based on the generalization gap that prevents the selection of architectures prone to overfitting. We present extensive experiments on the CIFAR-10, CIFAR-100 and ImageNet datasets and we perform search in two fundamentally different search spaces. We show that UNAS obtains the state-of-the-art average accuracy on all three datasets when compared to the architectures searched in the DARTS space. Moreover, we show that UNAS can find an efficient and accurate architecture in the ProxylessNAS search space, that outperforms existing MobileNetV2 based architectures.

In this paper, we compare the three most popular algorithms for hyperparameter optimization (Grid Search, Random Search, and Genetic Algorithm) and attempt to use them for neural architecture search (NAS). We use these algorithms for building a convolutional neural network (search architecture). Experimental results on CIFAR-10 dataset further demonstrate the performance difference between compared algorithms. The comparison results are based on the execution time of the above algorithms and accuracy of the proposed models.

Recent studies in using deep learning to solve the Travelling Salesman Problem (TSP) focus on construction heuristics, the solution of which may still be far from optimal-ity. To improve solution quality, additional procedures such as sampling or beam search are required. However, they are still based on the same construction policy, which is less effective in refining a solution. In this paper, we propose to directly learn the improvement heuristics for solving TSP based on deep reinforcement learning. We first present a reinforcement learning formulation for the improvement heuristic, where the policy guides selection of the next solution. Then, we propose a deep architecture as the policy network based on self-attention. Extensive experiments show that, improvement policies learned by our approach yield better results than state-of-the-art methods, even from random initial solutions. Moreover, the learned policies are more effective than the traditional handcrafted ones, and robust to different initial solutions with either high or poor quality. 1 Introduction The Travelling Salesman Problem (TSP) is a typical combinatorial optimization problem that has extensive applications in the real world. The problem statement is straightforward: given a set of locations, find the salesman a shortest tour that traverses each location exactly once and returns to the original one. Although having been widely studied for decades, achieving satisfactory performance is still challenging due to its NPhard complexity.

UAV enabled communications and networking can enhance wireless connectivity and support emerging services. However, this would require system-level understanding to modify and extend the existing terrestrial network infrastructure. In this paper, we integrate UAVs both as user equipment and base stations into existing LTE-Advanced heterogeneous network (HetNet) and provide system-level insights of this three-tier LTE-Advanced air-ground HetNet (AG-HetNet). This AG-HetNet leverages cell range expansion (CRE), ICIC, 3D beamforming, and enhanced support for UAVs. Using system-level understanding and through brute-force technique and heuristics algorithms, we evaluate the performance of AG-HetNet in terms of fifth percentile spectral efficiency (5pSE) and coverage probability. We compare 5pSE and coverage probability, when aerial base-stations (UABS) are deployed on a fixed hexagonal grid and when their locations are optimized using genetic algorithm (GA) and elitist harmony search algorithm based on genetic algorithm (eHSGA). Our simulation results show the heuristic algorithms outperform the brute-force technique and achieve better peak values of coverage probability and 5pSE. Simulation results also show that trade-off exists between peak values and computation time when using heuristic algorithms. Furthermore, the three-tier hierarchical structuring of FeICIC provides considerably better 5pSE and coverage probability than eICIC.

For the running example in Figure 1, this abstraction would replace the application-specific identifiers triangle and EQUILATERAL with generic placeholders, such as VAR1 and VAR2. After this abstraction, both approaches use an RNN-based sequence-to-sequence network that predicts how to modify the abstracted code. Given the increasing interest in learning-based approaches toward software engineering problems, we will likely see more progress on learning-based repair in the coming years. Key challenges toward effective solutions include finding an appropriate representation of source code changes and obtaining large amounts of high-quality human patches as training data.

Search methods, such as Genetic Algorithms and Simulated Annealing are typically used to achieve the required scalability in challenging problems for which it is hard to find optimal, or even just "good enough' solutions. The majority of these methods involve steps where the state of the algorithm is modified in some way to escape a local optimum. The aim is to avoid premature convergence, which is when the search method converges (usually very early in the search) to a local optimum of poor quality [1, 2]. Previous research has shown that the performance of search strategies is affected by the structure of the fitness landscape [3, 4]. A fitness landscape is defined by three components: i) the search space, which is the set of all candidate solutions, ii) the fitness function, which assigns a fitness value to each solution, and the neighbourhood operator, which defines how solutions are connected, and as a results how the search strategy can traverse the landscape.

There are four optimization algorithms to try. You can run a simple random search over the parameters. Nothing fancy here but it is useful to have this option within the same API to compare if needed. Both of those methods as well as the one in the next section are examples of Bayesian Hyperparameter Optimization also known as Sequential Model-Based Optimization SMBO. The idea behind this approach is to estimate the user-defined objective function with the random forest, extra trees, or gradient boosted trees regressor.

Recent superhuman results in games have largely been achieved in a variety of zero-sum settings, such as Go and Poker, in which agents need to compete against others. However, just like humans, real-world AI systems have to coordinate and communicate with other agents in cooperative partially observable environments as well. These settings commonly require participants to both interpret the actions of others and to act in a way that is informative when being interpreted. Those abilities are typically summarized as theory of mind and are seen as crucial for social interactions. In this paper we propose two different search techniques that can be applied to improve an arbitrary agreed-upon policy in a cooperative partially observable game. The first one, single-agent search, effectively converts the problem into a single agent setting by making all but one of the agents play according to the agreed-upon policy. In contrast, in multi-agent search all agents carry out the same common-knowledge search procedure whenever doing so is computationally feasible, and fall back to playing according to the agreed-upon policy otherwise. We prove that these search procedures are theoretically guaranteed to at least maintain the original performance of the agreed-upon policy (up to a bounded approximation error). In the benchmark challenge problem of Hanabi, our search technique greatly improves the performance of every agent we tested and when applied to a policy trained using RL achieves a new state-of-the-art score of 24.61 / 25 in the game, compared to a previous-best of 24.08 / 25. Introduction Real-world situations such as driving require humans to coordinate with others in a partially-observable environment with limited communication. In such environments, humans have a mental model of how other agents will behave in different situations (theory of mind). This model allows them to change their beliefs about the world based on why they think an agent acted as they did, as well as predict how their own actions will affect others' future behavior. Together, these capabilities allow humans to search for a good action to take while accounting for the behavior of others.

We introduce a novel keyword-aware influential community query KICQ that finds the most influential communities from an attributed graph, where an influential community is defined as a closely connected group of vertices having some dominance over other groups of vertices with the expertise (a set of keywords) matching with the query terms (words or phrases). We first design the KICQ that facilitates users to issue an influential CS query intuitively by using a set of query terms, and predicates (AND or OR). In this context, we propose a novel word-embedding based similarity model that enables semantic community search, which substantially alleviates the limitations of exact keyword based community search. Next, we propose a new influence measure for a community that considers both the cohesiveness and influence of the community and eliminates the need for specifying values of internal parameters of a network. Finally, we propose two efficient algorithms for searching influential communities in large attributed graphs. We present detailed experiments and a case study to demonstrate the effectiveness and efficiency of the proposed approaches.