A run-to-run controller for a chemical mechanical planarization process using least squares generative adversarial networks

Achieving high processing quality for chemical mechanical planarization (CMP) in semiconductor manufacturing is difficult due to the distinct process variations associated with this method, such as drift and shift. Run-to-run control aims to maintain the targeted process quality by reducing the effect of process variations. The goal of controller learning is to infer an underlying output–input reverse mapping based on input–output samples considering the process variations. Existing controllers learn reverse mapping by minimizing the total mapping error for sample data. However, this approach often fails to generate inputs for unseen target outputs because conditional input distributions on target outputs are not captured in the learning. In this study, we propose a controller based on a least squares generative adversarial network (LSGAN) that can capture the input distributions. GANs are deep-learning architectures composed of two neural nets: a generator and a discriminator. In the proposed model, the generator attempts to produce fake input distributions that are similar to the real input distributions considering the process variation features extracted using convolutional layers, while the discriminator attempts to detect the fake distributions. Competition in this game drives both networks to improve their performance until the generated input distributions are indistinguishable from the real distributions. An experiment using the data obtained from a work-site CMP tool verified that the proposed model outperformed the comparison models in terms of control accuracy and computation time.

     

Dynamic slack allocation algorithms for energy minimization on parallel machines

We explore novel algorithms for DVS (Dynamic Voltage Scaling) based energy minimization of DAG (Directed Acyclic Graph) based applications on parallel and distributed machines in dynamic environments. Static DVS algorithms for DAG execution use the estimated execution time. The estimated time in practice is overestimated or underestimated. Therefore, many tasks may be completed earlier or later than expected during the actual execution. For overestimation, the extra available slack can be added to future tasks so that energy requirements can be reduced. For underestimation, the increased time may cause the application to miss the deadline. Slack can be reduced for future tasks to reduce the possibility of not missing the deadline. In this paper, we present novel dynamic scheduling algorithms for reallocating the slack for future tasks to reduce energy and/or satisfy deadline constraints. Experimental results show that our algorithms are comparable to static algorithms applied at runtime in terms of energy minimization and deadline satisfaction, but require considerably smaller computational overhead.     

Vision-based arm gesture recognition for a long-range human–robot interaction

This paper proposes a vision-based human arm gesture recognition method for human–robot interaction, particularly at a long distance where speech information is not available. We define four meaningful arm gestures for a long-range interaction. The proposed method is capable of recognizing the defined gestures only with 320×240 pixel-sized low-resolution input images captured from a single camera at a long distance, approximately five meters from the camera. In addition, the system differentiates the target gestures from the users’ normal actions that occur in daily life without any constraints. For human detection at a long distance, the proposed approach combines results from mean-shift color tracking, short- and long-range face detection, and omega shape detection. The system then detects arm blocks using a background subtraction method with a background updating module and recognizes the target gestures based on information about the region, periodical motion, and shape of the arm blocks. From experiments using a large realistic database, a recognition rate of 97.235% is achieved, which is a sufficiently practical level for various pervasive and ubiquitous applications based on human gestures.     

Tiny and Blurred Face Alignment for Long Distance Face Recognition

Applying face alignment after face detection exerts a heavy influence on face recognition. Many researchers have recently investigated face alignment using databases collected from images taken at close distances and with low magnification. However, in the cases of home‐service robots, captured images generally are of low resolution and low quality. Therefore, previous face alignment research, such as eye detection, is not appropriate for robot environments. The main purpose of this paper is to provide a new and effective approach in the alignment of small and blurred faces. We propose a face alignment method using the confidence value of Real‐AdaBoost with a modified census transform feature. We also evaluate the face recognition system to compare the proposed face alignment module with those of other systems. Experimental results show that the proposed method has a high recognition rate, higher than face alignment methods using a manually‐marked eye position.     

3D Nanoprinting of Perovskites

As competing with the established silicon technology, organic–inorganic metal halide perovskites are continually gaining ground in optoelectronics due to their excellent material properties and low‐cost production. The ability to have control over their shape, as well as composition and crystallinity, is indispensable for practical materialization. Many sophisticated nanofabrication methods have been devised to shape perovskites; however, they are still limited to in‐plane, low‐aspect‐ratio, and simple forms. This is in stark contrast with the demands of modern optoelectronics with freeform circuitry and high integration density. Here, a nanoprecision 3D printing is developed for organic–inorganic metal halide perovskites. The method is based on guiding evaporation‐induced perovskite crystallization in mid‐air using a femtoliter ink meniscus formed on a nanopipette, resulting in freestanding 3D perovskite nanostructures with a preferred crystal orientation. Stretching the ink meniscus with a pulling process enables on‐demand control of the nanostructure's diameter and hollowness, leading to an unprecedented tubular‐solid transition. With varying the pulling direction, a layer‐by‐layer stacking of perovskite nanostructures is successfully demonstrated with programmed shapes and positions, a primary step for additive manufacturing. It is expected that the method has the potential to create freeform perovskite nanostructures for customized optoelectronics.     

Clausius Normalized Field‐Based Stereo Matching for Uncalibrated Image Sequences

We propose a homology between thermodynamic systems and images for the treatment of time‐varying imagery. A physical system colder than its surroundings absorbs heat from the surroundings. Furthermore, the absorbed heat increases the entropy of the system, which is closely related to its disorder as given by the definition of Clausius and Boltzmann. Because pixels of an image are viewed as a state of lattice‐like molecules in a thermodynamic system, the task of reckoning the entropy variations of pixels is similar to estimating their degrees of disorder. We apply this homology to the uncalibrated stereo matching problem. The absence of calibrations alleviates user efforts to install stereo cameras and enables users to freely modify the composition of the cameras. The proposed method is also robust to differences in brightness, white balancing, and even focusing between stereo image pairs. These peculiarities enable users to estimate the depths of interesting objects in practical applications without much effort in order to set and maintain a stereo vision setup. Users can consequently utilize two webcams as a stereo camera.     

On the adaptive real-time detection of fast-propagating network worms

We present two light-weight worm detection algorithms that offer significant advantages over fixed-threshold methods. The first algorithm, rate-based sequential hypothesis testing (RBS), aims at the large class of worms that attempts to quickly propagate, thus exhibiting abnormal levels of the rate at which hosts initiate connections to new destinations. The foundation of RBS derives from the theory of sequential hypothesis testing, the use of which for detecting randomly scanning hosts was first introduced by our previous work developing TRW (Jung et al. in Proceedings of the IEEE Symposium on Security and Privacy, 9–12 May 2004). The sequential hypothesis testing methodology enables us to engineer detectors to meet specific targets for false-positive and false-negative rates, rather than triggering when fixed thresholds are crossed. In this sense, the detectors that we introduce are truly adaptive. We then introduce RBS+TRW, an algorithm that combines fan-out rate (RBS) and probability of failure (TRW) of connections to new destinations. RBS+TRW provides a unified framework that at one end acts as pure RBS and at the other end as pure TRW. Selecting an operating point that includes both mechanisms extends RBS’s power in detecting worms that scan randomly selected IP addresses. Using four traces from three qualitatively different sites, we evaluate RBS and RBS+TRW in terms of false positives, false negatives, and detection speed, finding that RBS+TRW provides good detection of actual Code Red worm outbreaks that we caught in our trace as well as internal Web crawlers that we use as proxies for targeting worms. In doing so, RBS+TRW generates fewer than one false alarm per hour for wide range of parameter choices.     

DNS performance and the effectiveness of caching

This paper presents a detailed analysis of traces of domain name system (DNS) and associated TCP traffic collected on the Internet links of the MIT Laboratory for Computer Science and the Korea Advanced Institute of Science and Technology (KAIST). The first part of the analysis details how clients at these institutions interact with the wide-area domain name system, focusing on client-perceived performance and the prevalence of failures and errors. The second part evaluates the effectiveness of DNS caching. In the most recent MIT trace, 23% of lookups receive no answer; these lookups account for more than half of all traced DNS packets since query packets are retransmitted overly persistently. About 13% of all lookups result in an answer that indicates an error condition. Many of these errors appear to be caused by missing inverse (IP-to-name) mappings or NS records that point to nonexistent or inappropriate hosts. 27% of the queries sent to the root name servers result in such errors. The paper also presents the results of trace-driven simulations that explore the effect of varying time-to-live (TTL) and varying degrees of cache sharing on DNS cache hit rates. Due to the heavy-tailed nature of name accesses, reducing the TTL of address (A) records to as low as a few hundred seconds has little adverse effect on hit rates, and little benefit is obtained from sharing a forwarding DNS cache among more than 10 or 20 clients. These results suggest that client latency is not as dependent on aggressive caching as is commonly believed, and that the widespread use of dynamic low-TTL A-record bindings should not greatly increase DNS related wide-area network traffic.