Learning to learn: From smart machines to intelligent machines

Since its birth, more than five decades ago, one of the biggest challenges of artificial intelligence remained the building of intelligent machines. Despite amazing advancements, we are still far from having machines that reach human intelligence level. The current paper tries to offer a possible explanation of this situation. For this purpose, we make a review of different learning strategies and context types that are involved in the learning process. We also present the results of a study on cognitive development applied to the problem of face recognition for social robotics.     

Bend Passwords: using gestures to authenticate on flexible devices

Upcoming mobile devices will have flexible displays, allowing us to explore alternate forms of user authentication. On flexible displays, users can interact with the device by deforming the surface of the display through bending. In this paper, we present Bend Passwords, a new type of user authentication that uses bend gestures as its input modality. We ran three user studies to evaluate the usability and security of Bend Passwords and compared it to PINs on a mobile phone. Our first two studies evaluated the creation and memorability of user-chosen and system-assigned passwords. The third study looked at the security problem of shoulder-surfing passwords on mobile devices. Our results show that bend passwords are a promising authentication mechanism for flexible display devices. We provide eight design recommendations for implementing Bend Passwords on flexible display devices.     

Understanding hand gestures using approximate graph matching

We live in a society that depends on high-tech devices for assistance with everyday tasks, including everything from transportation to health care, communication, and entertainment. Tedious tactile input interfaces to these devices result in inefficient use of our time. Appropriate use of natural hand gestures will result in more efficient communication if the underlying meaning is understood. Overcoming natural hand gesture understanding challenges is vital to meet the needs of these increasingly pervasive devices in our every day lives. This work presents a graph-based approach to understand the meaning of hand gestures by associating dynamic hand gestures with known concepts and relevant knowledge. Conceptual-level processing is emphasized to robustly handle noise and ambiguity introduced during generation, data acquisition, and low-level recognition. A simple recognition stage is used to help relax scalability limitations of conventional stochastic language models. Experimental results show that this graph-based approach to hand gesture understanding is able to successfully understand the meaning of ambiguous sets of phrases consisting of three to five hand gestures. The presented approximate graph-matching technique to understand human hand gestures supports practical and efficient communication of complex intent to the increasingly pervasive high-tech devices in our society.     

A VE framework to study visual perception and action

In the real world, vision operates in harmony with self-motion yielding the observer to unambiguous perception of the three-dimensional (3D) space. In laboratory conditions, because of technical difficulties, researchers studying 3D perception have often preferred to use the substitute of a stationary observer, somehow neglecting aspects of the action-perception cycle. Recent results in visual psychophysics have proved that self-motion and visual processes interact, leading the moving observer to interpret a 3D virtual scene differently from a stationary observer. In this paper we describe a virtual environment (VE) framework which presents very interesting characteristics for designing experiments in visual perception during action. These characteristics arise in a number of ways from the design of a unique motion capture device. First, its accuracy and the minimal latency in position measurement; second, its ease of use and the adaptability to different display interfaces. Such a VE framework enables the experimenter to recreate stimulation conditions characterised by a degree of sensory coherence typical of the real world. Moreover, because of its accuracy and flexibility, the same device can be used as a measurement tool to perform elementary but essential calibration procedures. The VE framework has been used to conduct two studies which compare the perception of 3D variables of the environment in moving and in stationary observers under monocular vision. The first study concerns the perception of absolute distance, i.e. the distance separating an object and the observer. The second study refers to the perception of the orientation of a surface both in the absence and presence of conflicts between static and dynamic visual cues. In the two cases, the VE framework has enabled the design of optimal experimental conditions, permitting light to be shed on the role of action in 3D visual perception.     

From convex feasibility to convex constrained optimization using block action projection methods and underrelaxation

We describe the evolution of projection methods for solving convex feasibility problems to optimization methods when inconsistency arises, finally deriving from them, in a natural way, a general block method for convex constrained optimization. We present convergence results.     

Learning how to learn using a virtual learning environment for philosophy

Many students find theoretical subjects hard and challenging. Although they may pass modules in the subject, they often fail to attain a deeper, conceptual understanding. coMentor , a virtual learning environment on the WWW was developed to support such conceptual subjects by providing facilities for debate, discussion, groupwork, resource sharing and vicarious learning. The system was evaluated with undergraduates taking a philosophy module. Although there was no unequivocal evidence of gains in final assessment in those using the system, there was evidence of other forms of learning. Students learned from seeing each other's work, from having to 'write' down their ideas and share them with others and they adopted learning styles that were beneficial to learning a theoretical subject matter. Using a Learning Styles Inventory Scale, those using coMentor showed higher levels of deep learning and significantly higher levels of strategic learning than those who used the system little.     

Compression and recognition of dance gestures using a deformable model

In this paper, we aim for the recognition of a set of dance gestures from contemporary ballet. Our input data are motion trajectories followed by the joints of a dancing body provided by a motion-capture system. It is obvious that direct use of the original signals is unreliable and expensive. Therefore, we propose a suitable tool for non-uniform sub-sampling of spatiotemporal signals. The key to our approach is the use of a deformable model to provide a compact and efficient representation of motion trajectories. Our dance gesture recognition method involves a set of hidden Markov models (HMMs), each of them being related to a motion trajectory followed by the joints. The recognition of such movements is then achieved by matching the resulting gesture models with the input data via HMMs. We have validated our recognition system on 12 fundamental movements from contemporary ballet performed by four dancers. This revised version was published online in November 2004 with corrections to the section numbers. Ballet Atlantique Régine Chopinot.     

Utilizing gestures to improve sentence boundary detection

An accurate estimation of sentence units (SUs) in spontaneous speech is important for (1) helping listeners to better understand speech content and (2) supporting other natural language processing tasks that require sentence information. There has been much research on automatic SU detection; however, most previous studies have only used lexical and prosodic cues, but have not used nonverbal cues, e.g., gesture. Gestures play an important role in human conversations, including providing semantic content, expressing emotional status, and regulating conversational structure. Given the close relationship between gestures and speech, gestures may provide additional contributions to automatic SU detection. In this paper, we have investigated the use of gesture cues for enhancing the SU detection. Particularly, we have focused on: (1) collecting multimodal data resources involving gestures and SU events in human conversations, (2) analyzing the collected data sets to enrich our knowledge about co-occurrence of gestures and SUs, and (3) building statistical models for detecting SUs using speech and gestural cues. Our data analyses suggest that some gesture patterns influence a word boundary’s probability of being an SU. On the basis of the data analyses, a set of novel gestural features were proposed for SU detection. A combination of speech and gestural features was found to provide more accurate SU predictions than using only speech features in discriminative models. Findings in this paper support the view that human conversations are processes involving multimodal cues, and so they are more effectively modeled using information from both verbal and nonverbal channels.     

Novel side pose classification model of stretching gestures using three-layer LSTM

The Journal of Supercomputing - In recent years, low back pain rehabilitation exercises have been widely performed for spine-related illnesses. To facilitate rehabilitation exercises, pose-based...     

Implications of the ability to learn simple actions on the efficiency of the evolution of social behavior of agents

We investigate the effect of the ability to learn simple actions on the performance characteristics of the evolution of social behavior in agents situated in an inherently cooperative environment. Using a continuous predator–prey pursuit problem, we verified that relatively complex social behavior emerges from simple, implicit, locally defined, and thus robust and scalable interactions between the predator agents. Considering a distinct aspect of the phenomenon of emergence, we hypothesize that the ability of agents to learn how to perform simple, atomic acts of implicit interaction might facilitate the evolution of more complex behavior. The empirical results indicate that incorporation of the proposed approach to learning in genetic programming (employed as an algorithmic paradigm to evolve the social behavior of the agents) is associated with about a two-fold decrease in the computational effort of evolution.This work was presented, in part, at the 9th International Symposium on Artificial Life and Robotics, Oita, Japan, January 28–30, 2004     

基于手机加速度传感器的静态手势身份认证方法

针对智能手机上现有传统身份认证方法的缺点,提出一种基于人体生理震颤的静态手势身份认证方法。该方法在用户按特定静态姿势手持智能手机时,由手机内置的加速度传感器记录因持手机而产生的上肢肌肉震颤,经过信号预处理,得到信号的功率谱密度（ PSD）作为身份模板。通过动态时间规整（ DTW）算法计算待认证信号的PSD和模板之间的相似度,得到认证结果。实验结果表明：该方法的等错误率（ EER）和半错误率（ HTER）均在8%以下,验证了本方法的有效性。     

From the web of data to a world of action

This paper takes as its premise that the web is a place of action, not just information, and that the purpose of global data is to serve human needs. The paper presents several component technologies, which together work towards a vision where many small micro-applications can be threaded together using automated assistance to enable a unified and rich interaction. These technologies include data detector technology to enable any text to become a start point of semantic interaction; annotations for web-based services so that they can link data to potential actions; spreading activation over personal ontologies, to allow modelling of context; algorithms for automatically inferring ‘typing’ of web-form input data based on previous user inputs; and early work on inferring task structures from action traces. Some of these have already been integrated within an experimental web-based (extended) bookmarking tool, Snip!t, and a prototype desktop application On Time, and the paper discusses how the components could be more fully, yet more openly, linked in terms of both architecture and interaction. As well as contributing to the goal of an action and activity-focused web, the work also exposes a number of broader issues, theoretical, practical, social and economic, for the Semantic Web.     

Linking perception and action through motivation and affect

Research into cognitive architectures is described within a framework spanning major issues in artificial intelligence and cognitive science. Earlier work on motivation is extended with a cognitive model of reasoning which, together with an affective mechanism, enables consistent decision-making across a variety of cognitive and reactive processes. Cognition involves the control of behaviour within both external and internal environments. The control of behaviour is vital to an autonomous system as it acts to further its goals. Except in the most spartan of environments, the potential available information and associated combinatorics in a perception, cognition, and action sequence can tax even the most powerful agents. The affect magnitude concept solves some problems with BDI models, and allows for adaptive decision-making over a number of tasks in different domains. The cognitive and affective components are brought together using motivational constructs. The generic cognitive model can adapt to different environments and tasks as it makes use of motivational models to direct reactive and situated processes.     

Developing a hands-on activity using virtual reality to help students learn by doing

This study combined virtual reality (VR) technology, the 6E (Engage, Explore, Explain, Engineer, Enrich, and Evaluate) model, and STEM (Science, Technology, Engineering, and Mathematics) education to develop a hands-on activity aimed at helping students to achieve “learning by doing.” The participants were 162 tenth-grade students, divided into the Experimental Group (hands-on activity using VR technology) and the Control Group (hands-on activity via lectures). Using sequential analysis, this study investigated how the hands-on activity influenced the students' behavioral patterns in learning. The results showed that all of the students' learning performances and hands-on abilities were enhanced. Moreover, the students who used VR technology achieved both significantly better learning performances and hands-on abilities, indicating that VR might be able to help the students understand abstract scientific concepts and build mental models, which they used to internalize and organize knowledge structures. Furthermore, this study discovered that the students who learned using VR technology formed a cyclical learning pattern, starting with a group discussion (G), moving on to solving problems (S) and developing a product (D), and then going back to another group discussion. However, the students who learned via lectures produced a linear learning pattern in the order of G→S→D.     

False topographic perception phenomena observed with the satellite images of Moon's surface

False Topographic Perception Phenomena (FTPP) are commonly observed in images of rugged terrain on the surface of the Earth acquired by polar-orbiting satellites. It is interesting to note that due to the absence of atmosphere, vegetation and presence of numerous craters on its surface, the Moon illustrates very vividly and uniquely FTPP, which is less obvious with images of the Earth. In this article images of the Moon taken during the six missions of the NASA Apollo programme and from the Chandrayan-1 Indian satellite are used to demonstrate FTPP on the surface of the Moon. Numerous craters present on the surface of the Moon are perceived as hillock/plateau and vice versa. Combinations of various interrelated factors, for example topographic relief, observer position, Sun azimuth and elevation angles, are responsible for FTPP. Craters in the northern hemisphere of the Moon exhibit FTPP as the Moon's surface was imaged from southern side under equatorial illumination.     

Using hand gestures to control mobile spoken dialogue systems

Speech and hand gestures offer the most natural modalities for everyday human-to-human interaction. The availability of diverse spoken dialogue applications and the proliferation of accelerometers on consumer electronics allow the introduction of new interaction paradigms based on speech and gestures. Little attention has been paid, however, to the manipulation of spoken dialogue systems (SDS) through gestures. Situation-induced disabilities or real disabilities are determinant factors that motivate this type of interaction. In this paper, six concise and intuitively meaningful gestures are proposed that can be used to trigger the commands in any SDS. Using different machine learning techniques, a classification error for the gesture patterns of less than 5 % is achieved, and the proposed set of gestures is compared to ones proposed by users. Examining the social acceptability of the specific interaction scheme, high levels of acceptance for public use are encountered. An experiment was conducted comparing a button-enabled and a gesture-enabled interface, which showed that the latter imposes little additional mental and physical effort. Finally, results are provided after recruiting a male subject with spastic cerebral palsy, a blind female user, and an elderly female person.     

Ordered inference using observed confidence levels

Statistical inference on the ordering of the elements of a mean vector is an important issue in many applied problems, particularly in biostatistical applications. Some common ordering models include simple, tree and umbrella orderings. Many statistical methods have been developed to detect these orderings within the normal model, and outside the normal model using nonparametric methods. Estimates as well as confidence regions have also been developed for the mean vector under constraints imposed by these ordering models. In order to attempt to distinguish between ordered models, multiple testing procedures are usually required to control the overall error rate of the sequence of tests. This paper shows how observed confidence levels allow for the exploration of very general ordering models without the need for specialized asymptotic theory or multiple testing methods. The proposed methods are applied to several examples.     

Egomotion perception using visual tracking

The ability of a biological organism to visually track a perceptually significant feature in its environment has been argued to be an important feedback mechanism guiding locomotion. This paper analyzes the constraints available from the visual motion stimuli in the context of tracking. Our aim is to show that the act of tracking simplifies the decoding of egomotion parameters from motion stimuli. The constraints obtainable under tracking are utilized to analyze a possible egomotion decoding strategy for a binocular robot eye system, modeled after the human ocular tracking (smooth pursuit) mechanism. The main result of the paper is in the derivation of a closed-form solution of the egomotion parameters using feedback information concerning the movement of the tracking motors over time. The theoretical results are verified by experiments. We believe that the active tracking approach presented here is a more simple, practical, and manageable technique in a robot navigation setting, compared to passive methods.