We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

There are a huge number of features which are said to improve Convolutional Neural Network (CNN) accuracy. Practical testing of combinations of such features on large datasets, and theoretical justification of the result, is required. Some features operate on certain models exclusively and for certain problems exclusively, or only for small-scale datasets; while some features, such as batch-normalization and residual-connections, are applicable to the majority of models, tasks, and datasets. We assume that such universal features include Weighted-Residual-Connections (WRC), Cross-Stage-Partial-connections (CSP), Cross mini-Batch Normalization (CmBN), Self-adversarial-training (SAT) and Mish-activation. We use new features: WRC, CSP, CmBN, SAT, Mish activation, Mosaic data augmentation, CmBN, DropBlock regularization, and CIoU loss, and combine some of them to achieve state-of-the-art results: 43.5% AP (65.7% AP50) for the MS COCO dataset at a realtime speed of ~65 FPS on Tesla V100. Source code is at this https URL

YOLOv4: Optimal Speed and Accuracy of Object Detection

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

High-fidelity numerical simulation is suitable for analyzing the complex unsteady flow field dynamics of wind turbines. For a better understanding of these flow characteristics, the dynamic mode decomposition method can be used to carry out a reduced-order model study on the wakefield of wind turbines based on large-eddy simulations (LES) numerical simulation. In this paper, we abstract material dynamic information from the wakefield of the wind turbine by applying the sparsity-promoting dynamic mode decomposition (SPDMD) method, and the decomposition results are contrasted with the standard dynamic mode decomposition (DMD) method. Indicated that both mode decomposition methods can abstract the dynamic characteristics of wake and reveal the development and variation law of wind turbine wake. However, the frequency and spatial structure of the selected modes are different. For the purpose of demonstrating the extraction impact of the DMD/SPDMD method on the wakefield of wind turbines, DMD/SPDMD reduced-order models are established respectively. The result indicated that the relatively limited number of SPDMD modes is adequate to validly rehabilitate the wakefield of the unabridged wind turbine while standard DMD methods prerequisite more decomposition modes. Therefore, compared with the standard DMD method, the SPDMD method has strong robustness in mode selection, eliminates the feature information that contributes weakly to the flow, and has a smaller performance loss in the reconstruction of the wakefield of the wind turbine. The consumption of computing resources is greatly reduced.

Sparsity Promoting Dynamic Mode Decomposition for Data-Driven Modeling of Wind Turbine Wake

The task of annotating data into concise summaries poses a significant challenge across various domains, frequently requiring the allocation of significant time and specialized knowledge by human experts. Despite existing efforts to use large language models for annotation tasks, significant problems such as limited applicability to unlabeled data, the absence of self-supervised methods, and the lack of focus on complex structured data still persist. In this work, we propose a GPT self-supervision annotation method, which embodies a generating-recovering paradigm that leverages the one-shot learning capabilities of the Generative Pretrained Transformer (GPT). The proposed approach comprises a one-shot tuning phase followed by a generation phase. In the one-shot tuning phase, we sample a data from the support set as part of the prompt for GPT to generate a textual summary, which is then used to recover the original data. The alignment score between the recovered and original data serves as a self-supervision navigator to refine the process. In the generation stage, the optimally selected one-shot sample serves as a template in the prompt and is applied to generating summaries from challenging datasets. The annotation performance is evaluated by tuning several human feedback reward networks and by calculating alignment scores between original and recovered data at both sentence and structure levels. Our self-supervised annotation method consistently achieves competitive scores, convincingly demonstrating its robust strength in various data-to-summary annotation tasks.

GPT Self-Supervision for a Better Data Annotator

Study on the Tensile and Shear Behaviors of Selective Laser Melting Manufactured Ti6Al4V

For development of a simulator with a motion platform to generate an appropriate motion to reproduce the motion sense for the users, one of the most significant but disregarded complicated tasks is to build up a dynamic virtual motion model to reflect the motion of the simulated object in the corresponding physical world. Recently, a motion generation method based on motion blending technology was developed to alleviate the complication involved. It decomposes the simulated motion into a great number of parameterized motion blocks which are depicted by real motion data acquired from field tests and stored in a database. This paper proposes a streamlined motion blending technology suitable for a container crane simulator to further improve the current motion generation method based on the motion blending technology. Motion components, rather than motion blocks specially marked and stored in a database, are taken as the basic motion unit easily acquired through united analysis of crane dynamics and motion perception characteristics. They are then blended on demand to produce a one-stop model to directly act as the driving command of the motion platform without the need for a subsequent dedicated wash-out procedure. The calculation workload is greatly reduced and finally allows for achievement of higher fidelity of motion perceptions. Experiments are conducted to verify the effectiveness of the proposed streamlined motion blending technology for motion perception generation. Better training effect is found to be achieved due to more realistic simulation effects. The comprehensive training effectiveness index is enhanced from 54% to 82% once a motion simulation system developed using the proposed approach is introduced into the crane simulator.

/pdf/motion-generation-for-crane-simulators-using-streamlined-1osjskzs.pdf

Motion Generation for Crane Simulators Using Streamlined Motion Blending Technology

Searchable symmetric encryption (SSE) supports keyword search over outsourced symmetrically encrypted data. Dynamic searchable symmetric encryption (DSSE), a variant of SSE, further enables data updating. Most DSSE works with conjunctive keyword search primarily consider forward and backward privacy. Ideally, the server should only learn the result sets involving all keywords in the conjunction. However, existing schemes suffer from keyword pair result pattern (KPRP) leakage, revealing the partial result sets containing two of query keywords. We propose the first DSSE scheme to address aforementioned concerns that achieves strong privacy-preserving conjunctive keyword search. Specifically, our scheme can maintain forward and backward privacy and eliminate KPRP leakage, offering a higher level of security. The search complexity scales with the number of documents stored in the database in several existing schemes. However, the complexity of our scheme scales with the update frequency of the least frequent keyword in the conjunction, which is much smaller than the size of the entire database. Besides, we devise a least frequent keyword acquisition protocol to reduce frequent interactions between clients. Finally, we analyze the security of our scheme and evaluate its performance theoretically and experimentally. The results show that our scheme has strong privacy preservation and efficiency.

Chang Xu

Papers

Sparsity Promoting Dynamic Mode Decomposition for Data-Driven Modeling of Wind Turbine Wake

GPT Self-Supervision for a Better Data Annotator

Study on the Tensile and Shear Behaviors of Selective Laser Melting Manufactured Ti6Al4V

Motion Generation for Crane Simulators Using Streamlined Motion Blending Technology

Efficient Strong Privacy-Preserving Conjunctive Keyword Search Over Encrypted Cloud Data