Despite the emerging progress of integrating evolutionary computation into reinforcement learning, the absence of a high-performance platform endowing composability and massive parallelism causes nontrivial difficulties for research and applications related to asynchronous commercial games. Here, we introduce Lamarckian1—an open-source platform featuring support for evolutionary reinforcement learning scalable to distributed computing resources. To improve the training speed and data efficiency, Lamarckian adopts optimized communication methods and an asynchronous evolutionary reinforcement learning workflow. To meet the demand for an asynchronous interface by commercial games and various methods, Lamarckian tailors an asynchronous Markov Decision Process interface and designs an object-oriented software architecture with decoupled modules. In comparison with the state-of-the-art RLlib, we empirically demonstrate the unique advantages of Lamarckian on benchmark tests with up to 6000 CPU cores: 1) both the sampling efficiency and training speed are doubled when running proximal policy optimization (PPO) on Google football game; 2) the training speed is 13 times faster when running PBT+PPO onPonggame. Moreover, we also present two use cases: 1) how Lamarckian is applied to generating behavior-diverse game AI; 2) how Lamarckian is applied to game balancing tests for an asynchronous commercial game.

Emotion analysis of students plays an important role in teaching effect evaluation. To develop robust algorithms for emotion analysis of students, a database from real classrooms is required. However, most existing databases were collected from adults and constructed in laboratory settings. In this article, we present a manually-annotated facial action unit database from juveniles in real classrooms. Our database has three main characteristics: (1) it provides numerous education-related action units data from primary and high schools, complementing the vacancy of the publicly available educational action unit databases; (2) it contains 256,220 manually-annotated facial images of 1,796 juveniles, frame-by-frame annotated with 12 action units and 6-level intensities for each action unit; (3) it covers many challenges in the wild, including various head poses, low facial resolution, illuminations, and occlusions, supplementing action unit databases in the wild for research. The baselines for action unit detection and action unit intensity estimation are provided for future references. Especially, we apply the weighted balance loss to solve imbalanceswithinandbetweenlabels. Our database will be available to the research community:http://www.dlc.sjtu.edu.cn/rfau.

In this paper, we design an intelligent student behavior analysis system for recorded classrooms, which automatically detects hand-raising, standing, and sleeping behaviors of students. Detecting these behaviors is quite challenging mainly due to various scale behaviors, low resolution, and imbalanced behavior samples. To overcome the above-mentioned challenges, we first build a large-scale student behavior corpus from thirty schools, labeling these behaviors using bounding boxes frame-by-frame, which changes the behavior recognition problem into object detections. Then, we propose an improved Faster R-CNN, a classical object detection model, for student behavior analysis. Specifically, we first present a novel scale-aware detection head to overcome scale variations. Secondly, we propose a new feature fusion strategy to detect low-resolution behaviors while introduces little computation overhead. Thirdly, we utilize OHEM (Online Hard Example Mining) to alleviate severe class imbalances. Experiment results on our real corpus are increased by 3.4% mAP while maintaining a fast speed.

Generating diverse behaviors for game artificial intelligence (Game AI) has been long recognized as a challenging task in the game industry. Designing a Game AI with a satisfying behavioral characteristic (style) heavily depends on the domain knowledge and is hard to achieve manually. Deep reinforcement learning sheds light on advancing the automatic Game AI design. However, most of them focus on creating a superhuman Game AI, ignoring the importance of behavioral diversity in games. To bridge the gap, we introduce a new framework, named EMOGI, which can automatically generate desirable styles with almost no domain knowledge. More importantly, EMOGI succeeds in creating a range of diverse styles, providing behavior-diverse Game AIs. Evaluations on the Atari and real commercial games indicate that, compared to existing algorithms, EMOGI performs better in generating diverse behaviors and significantly improves the efficiency of Game AI design.

Students’ gestures, hand-raising, stand-up, and sleeping, indicates the engagement of students in classrooms and partially reflects teaching quality. Therefore, fast and automatically recognizing these gestures are of great importance. Due to limited computational resources in primary and secondary schools, we propose a real-time student behavior detector based on light-weight MobileNetV2-SSD to reduce the dependency of GPUs. Firstly, we build a large-scale corpus from real schools to capture various behavior gestures. Based on such a corpus, we transfer the gesture recognition task into object detections. Secondly, we design a multi-dimensional attention-based detector, named GestureDet, for real-time and accurate gesture analysis. The multi-dimensional attention mechanisms simultaneously consider all the dimensions of the training set, aiming to pay more attention to discriminative features and samples that are important for the final performance. Specifically, the spatial attention is constructed with stacked dilated convolution layers to generate a soft and learnable mask for re-weighting foreground and background features; the channel attention introduces the context modeling and squeeze-and-excitation module to focus on discriminative features; the batch attention discriminates important samples with a new designed reweight strategy. Experimental results demonstrate the effectiveness and versatility of GestureDet, which achieves 75.2% mAP on real student behavior dataset, and 74.5% on public PASCAL VOC dataset at 20fps on embedding device Nvidia Jetson TX2. Code will be made publicly available.

This paper proposes a novel method to detect sleep persons in real classroom scenes, which is useful for detecting the attention of students. There are several challenges for sleep gesture detection, including occlusion, various sleep gestures, interference terms with similar features like writing, and small sleep gesture targets. To solve these challenges, we first build a sleep gesture dataset from hundreds of real classes among schools in Shanghai. Second, to detect sleep gestures we propose a modified R-FCN integrated with feature pyramid and deformable convolution. Moreover, we design an efficient local multiscale testing algorithm to address small sleep gesture detection. Experiments on our sleep dataset have shown that our approach significantly outperforms the basic R-FCN and reaches 0.74 [email protected]. Especially, for small-size sleep gestures, our method gets an impressive improvement at 90% on our hard-samples dataset with just little additional time consumption. Besides, experiments on a large number of real video streams have proven our algorithm is applicable in real classroom scenes.

In this paper, we focus on dealing with problems of large-pose face alignment. Recently proposed heatmap-based algorithms have made promising performance on this problem. However, the traditional heatmap is constructed based on Gaussian model with fixed variance, which is inconsistent with the local shape of faces. In this paper, we propose a shape-aware heatmap to efficiently solve the problems of large-pose face alignment. Specifically, we design a novel heatmap based on Gaussian mixture model, where positions of several adjacent landmarks are utilized to construct different components. Thus the probability distribution is modified to fit the shape of the local region. The experimental results on Menpo-3D and AFLW2000-3D databases show that the proposed method outperforms the state-of-the-art algorithms.

Recently, time-unfolded recurrent neural network (RNN) based algorithms are successfully applied for fast sparse coding (SC) inference, such as LISTA and SLSTM. However, these methods do not properly exploit the historical information which is proved to speed up the convergence. In this paper, we propose a novel RNN-based SC inference framework with attention mechanism to efficiently incorporate the related historical information. The proposed framework consists of an attention network and a time-unfolded RNN, where the RNN generates the historical information and the attention network automatically determines the importance of these historical values for the current updating. The final sparse code is obtained by passing the context vectors generated from the attention network to a soft shrinkage function. Extensive experiments on convergence analysis and image classification tasks demonstrate that our network achieves impressive improvements on SC inference in terms of the quality of estimated sparse codes and the inference time. Moreover, the proposed network can be easily extended into a supervised version to further improve the classification accuracy.

Game testing has been long recognized as a notoriously challenging task, which mainly relies on manual playing and scripting based testing in game industry. Even until recently, automated game testing still remains to be largely untouched niche. A key challenge is that game testing often requires to play the game as a sequential decision process. A bug may only be triggered until completing certain difficult intermediate tasks, which requires a certain level of intelligence. The recent success of deep reinforcement learning (DRL) sheds light on advancing automated game testing, without human competitive intelligent support. However, the existing DRLs mostly focus on winning the game rather than game testing. To bridge the gap, in this paper, we first perform an in-depth analysis of 1349 real bugs from four real-world commercial game products. Based on this, we propose four oracles to support automated game testing, and further propose Wuji, an on-the-fly game testing framework, which leverages evolutionary algorithms, DRL and multi-objective optimization to perform automatic game testing. Wuji balances between winning the game and exploring the space of the game. Winning the game allows the agent to make progress in the game, while space exploration increases the possibility of discovering bugs. We conduct a large-scale evaluation on a simple game and two popular commercial games. The results demonstrate the effectiveness of Wuji in exploring space and detecting bugs. Moreover, Wuji found 3 previously unknown bugs, which have been confirmed by the developers, in the commercial games.

In this paper, we present a method (Action-Fusion) for human action recognition from depth maps and posture data using convolutional neural networks (CNNs). Two input descriptors are used for action representation. The first input is a depth motion image that accumulates consecutive depth maps of a human action, whilst the second input is a proposed moving joints descriptor which represents the motion of body joints over time. In order to maximize feature extraction for accurate action classification, three CNN channels are trained with different inputs. The first channel is trained with depth motion images (DMIs), the second channel is trained with both DMIs and moving joint descriptors together, and the third channel is trained with moving joint descriptors only. The action predictions generated from the three CNN channels are fused together for the final action classification. We propose several fusion score operations to maximize the score of the right action. The experiments show that the results of fusing the output of three channels are better than using one channel or fusing two channels only. Our proposed method was evaluated on three public datasets: 1) Microsoft action 3-D dataset (MSRAction3D); 2) University of Texas at Dallas-multimodal human action dataset; and 3) multimodal action dataset (MAD) dataset. The testing results indicate that the proposed approach outperforms most of existing state-of-the-art methods, such as histogram of oriented 4-D normals and Actionlet on MSRAction3D. Although MAD dataset contains a high number of actions (35 actions) compared to existing action RGB-D datasets, this paper surpasses a state-of-the-art method on the dataset by 6.84%.

In this paper, we propose two novel multi-dimensional tensor sparse coding (MDTSC) schemes using the t-linear combination. Based on the t-linear combination, the shifted versions of the bases are used for the data approximation, but without need to store them. Therefore, the dictionaries of the proposed schemes are more concise and the coefficients have richer physical explanations. Moreover, we propose an efficient alternating minimization algorithm, including the tensor coefficient learning and the tensor dictionary learning, to solve the proposed problems. For the tensor coefficient learning, we design a tensor-based fast iterative shrinkage algorithm. For the tensor dictionary learning, we first divide the problem into several nearly-independent subproblems in the frequency domain, and then utilize the Lagrange dual to further reduce the number of optimization variables. Experimental results on multi-dimensional signals denoising and reconstruction (3DTSC, 4DTSC, 5DTSC) show that the proposed algorithms are more efficient and outperform the state-of-the-art tensor-based sparse coding models.

In this paper, we propose an automatic hand-raiser recognition algorithm to show who raise their hands in real classroom scenarios, which is of great importance for further analyzing the learning states of individuals. To recognize the hand-raisers, we divide the hand-raiser recognition into three subproblems, including hand-raising detection, pose estimation, and matching the raised hands to students. Several challenges exist while dealing with the above-mentioned subproblems, such as low resolution of the back row for keypoints detection, the motion distortion caused by hand raising in pose estimation, and various complex situations for matching. To solve these challenges, we first adopt an improved R-FCN algorithm for hand-raising detection, whose effectiveness has been demonstrated. Secondly, we present a novel PAF-based pose estimation algorithm for detecting keypoints of human bodies. The proposed PAF adds scale search and modified weight metric to adapt to the real and complex scenarios. Specifically, scale search improves the detection effect at low resolution by pooling human characteristics in different sizes of pictures, and modified weight metric reasonably utilizes the directional vectors of possible limb connections to optimize the case of motion distortion. Thirdly, a heuristic matching strategy based on the location of hand-raising and keypoints information is proposed to recognize the hand-raisers. Experimental results on six teaching videos in real classrooms have demonstrated the efficiency of the proposed algorithm, and 83% recognition accuracy indicates the potential applications in real classrooms.

In this paper, we propose a novel low-rank tensor completion (LRTC) model under the circulant algebra for color image inpainting, which simultaneously preserves the low-rank structures of images, and also explore the local smooth and piecewise priors of the images in the spatial domain. First, color images are naturally represented by 3-order tensors which preserve the intrinsic structures of color images. Second, we preserve the low-rank structures of these tensors with tensor nuclear norm, which can simultaneously exploit the correlations among the spatial and channel domains. Third, we integrate an anisotropic total variation into our low-rank tensor completion model, which preserve the local smooth and piecewise priors of color images. Then, an efficient alternating direction method of multipliers (ADMM) is proposed to solve the resulting optimization problem. Experimental results on eight widely used color images demonstrate the effectiveness and superiority of the proposed algorithm.

This paper proposes a novel method for hand-raising detection in the real classroom environment. Different from traditional motion detection, the hand-raising detection is quite challenging in the real classroom due to complex scenarios, various gestures, and low resolutions. To solve these challenges, we first build up a large-scale hand-raising data set from thirty primary schools and middle schools of Shanghai, China. Then we propose an improved R-FCN to solve the above-mentioned challenges. Specifically, we first design an automatic detection templates algorithm for various gestures of hand-raising detection. Second, for better detection of the small-size hands, we present a feature pyramid to simultaneously capture the detail and highly semantic features. Incorporating these two strategies into a basic R-FCN architecture, our model achieves impressive results on real classroom scenarios. After a wide test, the accuracy of the hand-raising detection achieves 85% on average, which can satisfy the real application.

Recent low-rank based tensor completion (LRTC) algorithms have been successfully applied into color image inpainting. However, most of existing LRTC algorithms treat each dimension of tensors equally, which ignores the differences of the intrinsic structure correlations among dimensions. In this paper, we make a detailed analysis about the rank properties of each dimension and design a simple yet effective reweighted low-rank tensor completion model that truthfully capture the intrinsic structure correlations with reduced computational burden. Moreover, to capture the local smooth and piecewise priors of tensors, we integrate total variation into our model. Considering two formulations of LRTC, tensor unfolding and tensor decomposition, we propose corresponding two algorithms for color image recovery. Extensive experimental results on color image recovery show the efficiency and effectiveness of the proposed two algorithms against state-of-the-art competitors.

Action Unit (AU) detection is essential for facial expression analysis. However, most existing AU detection algorithms only focus on physical features, e.g., temporal feature and AU correlations, without considering various distributions of AUs, i.e., some AUs are quite less than others. In this work, we propose a novel cross-concat and temporal (CCT) neural network, which simultaneously consider physical features and the distribution differences. First, we design a cross-concat block (CCB) to adapt to the various distributions of AUs. CCB is based on the idea of skip connections since skip connections can reuse features from different layers and capture abundant features of AUs even with relatively small-size training samples. Second, LSTM layers are utilized to capture the temporal dependencies and multi-label learning is utilized for capturing AU correlations. Experimental results on three popular AU detection datasets, BP4D, DISFA, and GFT, show that the proposed algorithm outperforms the state-of-the-art ones.

Automatic Facial Action Unit (AU) detection from videos increases numerous interests over the past years due to its importance for analyzing facial expressions. Many proposed methods face challenges in detecting sparse face regions for different AUs, in the fusion of temporal dependency, and in learning multiple AUs simultaneously. In this paper, we propose a novel deep neural network architecture for AU detection to model above-mentioned challenges jointly. Firstly, to capture the region sparsity, we design a region pooling layer after a fully convolutional network to extract per-region features for each AU. Secondly, in order to integrate temporal dependency, Long Short Term Memory (LSTM) is stacked on the top of regional features. Finally, the regional features and outputs of LSTMs are utilized together to produce per-frame multi-label predictions. Experimental results on three large spontaneous AU datasets, BP4D, GFT and DISFA, have demonstrated our work outperforms state-of-the-art methods. On three datasets, our work has highest average F1 and AUC scores with an average F1 score improvement of 4.8% on BP4D, 12.7% on GFT and 14.3% on DISFA, and an average AUC score improvement of 27.4% on BP4D and 33.5% on DISFA.

Optical coherence tomography is an immersive technique for depth analysis of retinal layers. Automatic choroid layer segmentation is a challenging task because of the low contrast inputs. Existing methodologies carried choroid layer segmentation manually or semi‐automatically. The authors proposed automated choroid layer segmentation based on normalised cut algorithm, which aims at extracting the global impression of images and treats the segmentation as a graph partitioning problem. Due to the structure complexity of retinal and choroid layers, the authors employed a series of pre‐processing to make the cut more deterministic and accurate. The proposed method divided the image into several patches and ran the normalised cut algorithm on every patch separately. The aim was to avoid insignificant vertical cuts and focus on horizontal cutting. After processing every patch, the authors acquired a global cut on the original image by combining all the patches. Later the authors measured the choroidal thickness which is highly helpful in the diagnosis of several retinal diseases. The results were computed on a total of 525 images of 21 real patients. Experimental results showed that the mean relative error rate of the proposed method was around 0.4 when compared with the manual segmentation performed by the experts.

Sparse coding (SC) is a unsupervised learning scheme that has received an increasing amount of interests in recent years. However, conventional SC vectorizes the input images, which destructs the intrinsic spatial structures of the images. In this paper, we propose a novel graph regularized tensor sparse coding (GTSC) for image representation. GTSC preserves the local proximity of elementary structures in the image by adopting the newly proposed tubal-tensor representation. Simultaneously, it considers the intrinsic geometric properties by imposing graph regularization that has been successfully applied to uncover the geometric distribution for the image data. Moreover, the learned sparse representations by GTSC have better physical explanations as the key operation (i.e., circular convolution) in the tubal-tensor model preserves the shifting invariance property. Experimental results on image clustering demonstrate the effectiveness of the proposed scheme.

As the standard colour space used by printers, Cyan, Magenta, Yellow, Black (CMYK) colour model is a subtractive colour space used to describe the printing process. Existing CMYK conversion methods rely on static conversion table, which may not preserve the subtle visual structures of images, due to the local visual contrast loss caused by the static colour mapping. Therefore, the authors propose a novel dynamic Red, Green, Blue (RGB)‐to‐CMYK colour conversion, which utilises the weighted entropy to extract the pixels with filter response change dramatically. They obtain the image activity map by combining these pixels with high skin probability regions, and optimise the colour conversion of each pixel to ensure that the ink used for each pixel can be saved, while the visual contrast can be preserved with ink‐saving. In this way, their proposed technique can achieve dynamic CMYK colour conversion, in which the consumption of ink can be reduced without the loss of visual contrast. The experimental results have shown that their dynamic CMYK colour conversion saved 10–25% ink consumption compared with the static conversion method, while with high visual quality for the converted images.

Although many-objective optimisation can be simplified through reduction of redundant objectives, algorithms that perform this reduction still lack a convenient method of evaluation. In this paper, we address this deficiency by proposing a new method of evaluation, on the basis of changes in the Pareto-domination ratio after a reduction has occurred. Experimental results have shown that the proposed method can perform non- redundant objective set evaluation more accurately than existing evaluation methods, and also does not need the true Pareto front beforehand.

Public vehicle (PV) systems will be efficient traffic-management platforms in future smart cities, where PVs provide ridesharing trips with balanced QoS (quality of service). PV systems differ from traditional ridesharing due to that the paths and scheduling tasks are calculated by a server according to passengers' requests, and all PVs corporate with each other to achieve higher transportation efficiency. Path planning is the primary problem. The current path planning strategies become inefficient especially for traffic big data in cities of large population and urban area. To ensure real-time scheduling, we propose one efficient path planning strategy with balanced QoS (e.g., waiting time, detour) by restricting search area for each PV, so that a large number of computation is saved. Simulation results based on the Shanghai (China) urban road network show that, the computation can be reduced by 34% compared with the exhaustive search method since many requests violating QoS are excluded.

This paper advocates a new paradigm of transportation systems for future smart cities, namely, public vehicles (PVs), that provides dynamic ridesharing trips at requests. Passengers will enjoy more convenient and flexible transportation services with much less expense. In the PV system, both the number of vehicles and required parking spaces will be significantly reduced. There will be less traffic congestion, less energy consumption, and less pollution. In this paper, the concept, method, and algorithm for the PV system are described. The key issue of effectively implementing the PV system is to design efficient planning and scheduling algorithms. The PV-path problem is formulated, which is NP-complete. Then, a practical approach is proposed, which can serve people anywhere and anytime. The simulation results show that, to achieve the same performance (e.g., total time, waiting time, and travel time), the number of vehicles in the PV system can be reduced by around 90% and 57% compared with the conventional vehicle system and Uber Pool, respectively, and the total traveling distance can be reduced by 34% and 14%.

In this paper, we present a system that extracts the information implied by eye movements and use this information to analyze students' learning behavior. Our system uses a common webcam to capture students' facial image sequences when they are learning in front of monitors. We then process these images and establish sequences of changing location of iris, which represent the movements of eyes. With the eye movement sequences, we train a HMM classifier that can analyze their pattern and generate learning status for any given moment in the lesson. These statuses could help the computers to get a better understanding about the students' intention and behavior during online learning. The status sequences of those who view the same lesson could also be used as a reference for teaching quality assessment.

This paper introduces an optimization template library (OTL), a cross-platform C++ template library for multiobjective optimization. OTL has an object-oriented architecture, which allows that different modules can be arbitrarily combined with each other. Moreover, the C++ template technique is used to increase the flexibility of OTL. Meanwhile, generic programming is widely used in OTL, and the generic algorithms can be used to process different data structures. However, compared with C++, the Python script is more suitable for building the experimental platform. To ensure that all attributes of the experimental results can be fully maintained, a database is used to store the experimental data. Moreover, batch experiments can be easily defined in a set of configuration files; thus, the experiments can be executed automatically without human intervention. In addition, serial and various parallel execution modes are supported, and the user can easily switch the running mode to distributed computing to increase the computing speed. Finally, a highly customizable data visualization tool is created to play back the data sample stored in the database. From a series of comparative studies, the accuracy and running performance of OTL are verified by the statistical results.

Substantial technology advances have been made in areas of autonomous and connected vehicles, which opens a wide landscape for future transportation systems. We propose a new type of transportation system, Public Vehicle (PV) system, to provide effective, comfortable, and convenient service. The PV system is to improve the efficiency of current transportation systems, \eg, taxi system. Meanwhile, the design of such a system targets on significant reduction in energy consumption, traffic congestion, and provides solutions with affordable cost. The key issue of implementing an effective PV system is to design efficient scheduling algorithms. We formulate it as the PV Path (PVP) problem, and prove it is NP-Complete. Then we introduce a real time approach, which is based on solutions of the Traveling Salesman Problem (TSP) and it can serve people efficiently with lower costs. Our results show that to achieve the same performance (e.g., the total time: waiting and travel time), the number of vehicles can be reduced by 47%-69%, compared with taxis. The number of vehicles on roads is reduced, thus traffic congestion is relieved.

The demands of an increasingly knowledge based society and the advances in mobile phone technology are combining to spur the growth of mobile learning. However, for mobile learning to attain its full potential, it is essential to develop more advanced technologies that are tailored to the needs of this new learning environment. We present a non-invasive emotion-aware m-learning model based on multi-modal emotion detection, to deliver learning services that correspond to the students' needs, and to adapt a learning system according to a student's status, thereby enhancing the student's learning experience. The research presented in this paper is to augment our m-learning system in daily use at the School of Continuing Education of Shanghai Jiao Tong University, a blended-learning institution with 35.000 students.

Electric vehicle (EV) is a promising transportation with plenty of advantages, e.g., low carbon emission, high energy efficiency. However, it requires frequent and long time charging. In public charging stations, EVs spend long time on queuing especially during peak hours. Hence, it requires an efficient method to reduce the total charging time for EVs. We study the Electric Vehicle Charging-Scheduling (EVCS) problem in this paper. First we prove that EVCS is NP-Complete, which can be reduced from one Parallel Machine Scheduling (PMS) problem. Then two heuristic algorithms are proposed: the Earliest Start Time (EST) algorithm, and the Earliest Finish Time (EFT) algorithm. EST tries to advance the start charging time to get customers in service as early as possible, while EFT focuses on the possible finish charging time to get customers served as soon as possible. Finally simulations show that, the proposed algorithms outperform the classic greedy nearest scheduling algorithm: assign each EV to its nearest charging station, then choose the outlet where the fewest EVs are queuing. Typically, under our simulation settings, the average finish time and maximum finish time can be reduced by about one hour, and six hours respectively.

The last years have seen an increased usage of apps, applications that are downloaded from a centralized repository and run in a particular environment. However, smaller app environments face the major challenge of attracting sufficient awareness and usage by third-party developers and users to become sustainable. This is a typical chicken-and-egg problem. Without apps, few people will be using the environment. Few users lead to limited incentive for app developers. We analyze work done in a very similar domain, namely learning object repositories (LORs), to learn about factors that inhibit participation in app environments. We also present an approach to overcome these problems, which aims at lowering the barriers of participation for app production as much as possible. The framework was partly implemented and is being applied at a higher education institution, where teachers and teaching assistants created over a hundred of apps.

A large amount of educational content is available as lecture videos, which record teachers as they proceed through a course. Students watch these videos in different ways. They rewind, skip forward, watch some scenes repeatedly. This work investigates what can be learned by analyzing such viewing patterns. We show how to use machine learning techniques to analyze such data, and present the outcomes of an analysis of data collected from the interactions of 2992 students in 253 courses. The viewing pattern were put into relation to seven different variables, such as the final score of the student and the rating teachers received from students Our analysis shows that some variables, such as the teacher rating, were indeed predictable from the viewing patterns.

This paper presents an approach to quickly transform online resources into widgets. This approach has several advantages over previous solutions. Two small-scale evaluations indicate that this solution empowers the average teachers to become a widget author, without any programming knowledge. The teachers showed their enthusiasm to use this tool in the near future and believe that this authoring tool will help them both during the authoring and teaching process. There are still many features required to be added to improve our tool. For example, the automatic generation of metadata might improve usability and ease of use even more, especially when adding category information to the metadata to ease later reuse for existing widgets.

This paper presents a system, called E-supervisor that detects a student's state of attention and provides appropriate feedback. The system uses an eye detection based method and provides multi-mode feedback such as screen alerting, chair vibration, scent emitting and music playing to make participants aware whenever they are not paying attention. An evaluation shows that supervision by the E-Supervisor can improve learning and attention. Furthermore perceived usefulness is judged high by the evaluation participants, with the vibrant feedback being the preferred feedback modality.

Live screencasting involves encoding and streaming the screen content of a PC in real-time. Most existing H.264 rate control (RC) algorithms are designed for natural scenes and do not perform well with the quite different signal characteristics of screencasting. This paper proposes a content-adaptive H.264 RC scheme which classifies screen content as “slow-motion” phase or “fast-motion” phase on the fly, based on the temporal and spatial frame-area update pattern of recent frames. Then, for frames in “slow-motion” phase, which usually result from a presenter's GUI operations, a new RC algorithm named Frame Rate Adaptive-CQP (FRA-CQP) is applied, which puts priority on the quality of individual frame rather than the frame rate. For frames in “fast-motion” phase, which usually result from playing a movie during the presentation, the classical CRF+VBV RC algorithm is applied. Evaluation results show this adaptive RC scheme can regularly achieves higher subjective quality assessment score (0.7 to 1.6 points on a 1-5 scale) than existing algorithms while meeting the RC objective.

Online learners are individuals, and their learning abilities, knowledge, and learning performance differ substantially and are ever changing. These individual characteristics pose considerable challenges to online learning courses. In this paper, we propose an online course generation and evolution approach based on genetic algorithms to provide personalized learning. The courses generated consider not only the difficulty level of a concept and the time spent by an individual learner on the concept, but also the changing learning performance of the individual learner during the learning process. We present a layered topological sort algorithm, which converges towards an optimal solution while considering multiple objectives. Our general approach makes use of the stochastic convergence of genetic algorithms. Experimental results show that the proposed algorithm is superior to the free browsing learning mode typically enabled by online learning environments because of the precise selection of learning content relevant to the individual learner, which results in good learning performance.

Since learning in Boltzmann machines is typically quite slow, there is a need to restrict connections within hidden layers. However, theresulting states of hidden units exhibit statistical dependencies. Based on this observation, we propose using l1/l2 regularization upon the activation probabilities of hidden units in restricted Boltzmann machines to capture the local dependencies among hidden units. This regularization not only encourages hidden units of many groups to be inactive given observed data but also makes hidden units within a group compete with each other for modeling observed data. Thus, the l1/l2 regularization on RBMs yields sparsity at both the group and the hidden unit levels. We call RBMs trained with the regularizer sparse group RBMs (SGRBMs). The proposed SGRBMs are appliedto model patches of natural images, handwritten digits and OCR English letters. Then to emphasize that SGRBMs can learn more discriminative features we applied SGRBMs to pretrain deep networks for classification tasks. Furthermore, we illustrate the regularizer can also be applied to deep Boltzmann machines, which lead to sparse group deep Boltzmann machines. When adapted to the MNIST data set, a two-layer sparse group Boltzmann machine achieves an error rate of 0.84%, which is, to our knowledge, the best published result on the permutation-invariant version of the MNIST task.

Grid has been widely used in the field of evolutionary multi-objective optimization (EMO) due to its property combining convergence and diversity naturally. Most EMO algorithms of grid-based fitness perform well on problems with two or three objectives, but encounter difficulties in their scalability to many-objective optimization. This paper develops the potential of using grid technique to balance convergence and diversity in fitness for many-objective optimization problems. To strengthen selection pressure and refine comparison level, three hierarchical grid-based criterions are incorporated into fitness to establish a completer order among individuals. Moreover, an adaptive fitness penalty mechanism in environmental selection is employed to guarantee the diversity of archive memory. Based on an extensive comparative study with three other EMO algorithms, the proposed algorithm is found to be remarkably successful in finding well-converged and well-distributed solution set.

In China, the number of online learners who attend formal education has quadrupled in the last 5 years to 8.2 millions until the end of 2008. How can online teachers build and update web-courses for such a large numbers of online students - courses that ideally respect the different requirements of the students? In this paper we describe an automatic course generation System (ACGS) developed at the Shanghai Jiao Tong University (SJTU). In this ACGS, teachers can build web-based courses according to their own instructional plan and publish the web-based courses without requiring technological support. Teachers can also update the web-courses whenever the instructional objective changes. The web-course learning environment generated from this system takes into account pedagogical scenarios. The ease of use of our system is illustrated by its quickly increasing number of users. About 50 teachers in the SJTU School of Continuing Education (SOCE-SJTU) have developed 45 web-courses in only two months. These 45 web-courses have been provided to about 5800 online learners in the last 2009 autumn semester alone.

The term is used to describe applications that distinguish themselves from previous generations of software by a number of principles. Existing work shows that Web 2.0 applications can be successfully exploited for technology-enhance learning. However, in-depth analyses of the relationship between Web 2.0 technology on the one hand and teaching and learning on the other hand are still rare. In this article, we will analyze the technological principles of the Web 2.0 and describe their pedagogical implications on learning. We will furthermore show that Web 2.0 is not only well suited for learning but also for research on learning: the wealth of services that is available and their openness regarding API and data allow to assemble prototypes of technology-supported learning applications in amazingly small amount of time. These prototypes can be used to evaluate research hypotheses quickly. We will present two example prototypes and discuss the lessons we learned from building and using these prototypes.

With the fast development of mobile computing terminals and wireless network, mobile learning (ML) is becoming one of the most popular research and development areas in today's e-learning. Through convenient and flexible mobile terminals and network, researchers and developers strive to provide users with teaching and learning resources and services that can be accessible anytime anywhere. This paper describes the design and implementation of a mobile learning system that can deliver live broadcast of realtime classroom teaching to online students with mobile devices. The system allows students to customize their ways of receiving, based on when and where they are tuning into the broadcast. To increase interactivity between the instructor and online students in large classes, the system also supports short text-messaging and instant polls. Through these venues, students can ask questions and make suggestions in real time, and the instructor can address them immediately

In this paper, a novel object-based robust watermarking scheme against geometric attacks is proposed. The synchronization problem can be solved by object normalization. Binary logo watermark is embedded into the lower frequency bands of the DWT domain of the video object by adjusting the relation of the selected coefficient pairs. The theoretical analysis and experimental results show good robustness of proposed scheme against object manipulations and some affine transforms.

In this paper, we derive a general sufficient condition ensuring global exponential convergence of Cohen-Grossberg neural networks with time delays by constructing a novel Lyapunov functional and smartly estimating its derivative. The proposed condition is related to the convex combinations of the column-sum and the row-sum of the connection matrices and also relaxes the constraints on the network coefficients. Therefore, the proposed condition generalizes some previous results in the literature.

This paper analyzes the architecture of virtual experiment and discusses the possibility of building a standard interface for it. We also introduces an agent-based VE system which can easily integrate apparatus developed independently and through which a teacher can just focus on his own field even without any programming technique. This model may greatly eliminate many problems involved in a software engineer and make a distributed experiment possible.