Search

The AlphaZero algorithm achieved superhuman levels of play in chess, shogi, and Go by learning without domain-specific knowledge except for game rules. This paper targets stochastic games and investigates whether AlphaZero can learn theoretical values and optimal play. Since the theoretical values of stochastic games are expected win rates, not a simple win, loss, or draw, it is worth investigating the ability of AlphaZero to approximate expected win rates of positions. This paper also thoroughly studies how AlphaZero is influenced by hyper-parameters and some implementation details. The analyses are mainly based on AlphaZero learning with lookup tables. Deep neural networks (DNNs) like the ones in the original AlphaZero are also experimented and compared. The tested stochastic games include reduced and stronglysolved variants of Chinese dark chess and EinStein würfelt nicht!. The experiments showed that AlphaZero could learn policies that play almost optimally against the optimal player and could learn values accurately. In more detail, such good results were achieved by different hyper-parameter settings in a wide range, though it was observed that games on larger scales tended to have a little narrower range of proper hyper-parameters. In addition, the results of learning with DNNs were similar to lookup tables.

This paper evaluates speech emotion and naturalness recognitions by utilizing deep learning models with multitask learning and single-task learning approaches. The emotion model accommodates valence, arousal, and dominance attributes known as dimensional emotion. The naturalness ratings are labeled on a five-point scale as dimensional emotion. Multitask learning predicts both dimensional emotion (as the main task) and naturalness scores (as an auxiliary task) simultaneously. The single-task learning predicts either dimensional emotion (valence, arousal, and dominance) or naturalness score independently. The results with multitask learning show improvement from previous studies on single-task learning for both dimensional emotion recognition and naturalness predictions. Within this study, single-task learning still shows superiority over multitask learning for naturalness recognition. The scatter plots of emotion and naturalness prediction scores against the true labels in multitask learning exhibit the lack of the model; it fails to predict the low and extremely high scores. The low score of naturalness prediction in this study is possibly due to a low number of samples of unnatural speech samples since the MSP-IMPROV dataset promotes the naturalness of speech. The finding that jointly predicting naturalness with emotion helps improve the performance of emotion recognition may be embodied in the emotion recognition model in future work.

Privacy protection has attracted increasing attention, and privacy concerns often prevent flexible data utilization. In most industries, data are distributed across multiple organizations due to privacy concerns. Federated learning (FL), which enables cross-organizational machine learning by communicating statistical information, is a state-of-the-art technology that is used to solve this problem. However, for gradient boosting decision tree (GBDT) in FL, balancing communication efficiency and security while maintaining sufficient accuracy remains an unresolved problem. In this paper, we propose an FL scheme for GBDT, i.e., efficient FL for GBDT (eFL-Boost), which minimizes accuracy loss, communication costs, and information leakage. The proposed scheme focuses on appropriate allocation of local computation (performed individually by each organization) and global computation (performed cooperatively by all organizations) when updating a model. It is known that tree structures incur high communication costs for global computation, whereas leaf weights do not require such costs and are expected to contribute relatively more to accuracy. Thus, in the proposed eFL-Boost, a tree structure is determined locally at one of the organizations, and leaf weights are calculated globally by aggregating the local gradients of all organizations. Specifically, eFL-Boost requires only three communications per update, and only statistical information that has low privacy risk is leaked to other organizations. Through performance evaluation on public data sets (ROC AUC, Log loss, and F1-score are used as metrics), the proposed eFL-Boost outperforms existing schemes that incur low communication costs and was comparable to a scheme that offers no privacy protection.

Three-dimensional (3D) sensor networks using multiple light-detection-and-ranging (LIDAR) sensors are good for smart monitoring of spots, such as intersections, with high potential risk of road-traffic accidents. The image sensors must share the strictly limited computation capacity of an edge computer. To have the computation speeds required from real-time applications, the system must have a short computation delay while maintaining the quality of the output, e.g., the accuracy of the object detection. This paper proposes a spatial-importance-based computation scheme that can be implemented on an edge computer of image-sensor networks composed of 3D sensors. The scheme considers regions where objects exist as more likely to be ones of higher spatial importance. It processes point-cloud data from each region according to the spatial importance of that region. By prioritizing regions with high spatial importance, it shortens the computation delay involved in the object detection. A point-cloud dataset obtained by a moving car equipped with a LIDAR unit was used to numerically evaluate the proposed scheme. The results indicate that the scheme shortens the delay in object detection.

In cooperative relaying, intermediate stations are required to enhance the end-to-end transmission performance. The performance of the cooperative relaying scheme has been investigated theoretically and via computer simulations. However, cooperative relaying using transmit diversity techniques in actual environments has not been investigated thus far. This paper presents an experimental system for distributed cooperative relaying using space-time block code and evaluations of its transmission performances in real propagation channels. To this end, four wireless stations-specifically, one source, two relays, and one destination-were developed using analog transceivers and field-programmable gate arrays for real-time digital signal processing. Sample timing and frequency synchronizations among the four wireless stations were established by using the received signals as a reference. The end-to-end error performance of distributed cooperative relaying was compared to those of noncooperative relaying schemes, and the performances of three relaying schemes were evaluated quasisimultaneously in terms of their cumulative distribution functions of the bit-error ratios (BERs). The experimental results indicated that the BER performance of the two-hop distributed cooperative relaying scheme was substantially superior to those of noncooperative two-hop relaying schemes, including a route diversity scheme.

This paper addresses bandwidth allocation to multiple light detection and ranging (LIDAR) sensors for smart monitoring, which a limited communication capacity is available to transmit a large volume of point-cloud data from the sensors to an edge server in real time. To deal with the limited capacity of the communication channel, we propose a bandwidth-allocation scheme that assigns multiple point-cloud compression formats to each LIDAR sensor in accordance with the spatial importance of the point-cloud data transmitted by the sensor. Spatial importance is determined by estimating how objects, such as cars, trucks, bikes, and pedestrians, are likely to exist since regions where objects are more likely to exist are more useful for smart monitoring. A numerical study using a real point-cloud dataset obtained at an intersection indicates that the proposed scheme is superior to the benchmarks in terms of the distributions of data volumes among LIDAR sensors and quality of point-cloud data received by the edge server.

The cornea in the human eye reflects incoming environmental light, which means we can obtain information about the surrounding environment from the corneal reflection in facial images. In recent years, as the quality of consumer cameras increases, this has caused privacy concerns in terms of identifying the people around the subject or where the photo is taken. This paper investigates the security risk of eye corneal reflection images: specifically, visual place recognition from eye reflection images. First, we constructed two datasets containing pairs of scene and corneal reflection images. The first dataset is taken in a virtual environment. We showed pre-captured scene images in a 180-degree surrounding display system and took corneal reflections from subjects. The second dataset is taken in an outdoor environment. We developed several visual place recognition algorithms, including CNN-based image descriptors featuring a naive Siamese network and AFD-Net combined with entire image feature representations including VLAD and NetVLAD, and compared the results. We found that AFD-Net+VLAD performed the best and was able to accurately determine the scene in 73.08% of the top-five candidate scenes. These results demonstrate the potential to estimate the location at which a facial picture was taken, which simultaneously leads to a) positive applications such as the localization of a robot while conversing with persons and b) negative scenarios including the security risk of uploading facial images to the public.

For densely deployed wireless local area networks (WLANs), this paper proposes a deep reinforcement learning-based channel allocation scheme that enables the efficient use of experience. The central idea is that an objective function is modeled relative to communication quality as a parametric function of a pair of observed topologies and channels. This is because communication quality in WLANs is significantly influenced by the carrier sensing relationship between access points. The features of the proposed scheme can be summarized by two points. First, we adopt graph convolutional layers in the model to extract the features of the channel vectors with topology information, which is the adjacency matrix of the graph dependent on the carrier sensing relationships. Second, we filter experiences to reduce the duplication of data for learning, which can often adversely influence the generalization performance. Because fixed experiences tend to be repeatedly observed in WLAN channel allocation problems, the duplication of experiences must be avoided. The simulation results demonstrate that the proposed method enables the allocation of channels in densely deployed WLANs such that the system throughput increases. Moreover, improved channel allocation, compared to other existing methods, is achieved in terms of the system throughput. Furthermore, compared to the immediate reward maximization method, the proposed method successfully achieves greater reward channel allocation or realizes the optimal channel allocation while reducing the number of changes.