In a context of digital transformation, the widespread use of algorithmic systems is a given in modem society. In spite of their growing popularity, sorne of these systems often exhibit a puzzling behaviour resulting ...
详细信息
In a context of digital transformation, the widespread use of algorithmic systems is a given in modem society. In spite of their growing popularity, sorne of these systems often exhibit a puzzling behaviour resulting in a lack of understanding and comprehension of their inner workings by humans (i. e. the so-called 'black-box problem'). Focused on algorithmic opacity in the case of deep machine leaming models, this qualitative explanation-driven research takes the form of a literature review followed by conceptual developments aimed at explaining algorithmic opacity in sociological terms. I define algorithmic opacity as the lack of human understanding of algorithmic logic and its execution by a machine. From a sociological perspective, the overall mechanism involves the presence of (a) social actors, their properties and actions; (b) their decision-making processes; ( c) their interpretation of the results of an execution of an algorithm by a machine; and ( d) their specifie social context. algorithmic opacity then may be understood as a result of the accumulation of trade-offs and other by-products of human decision-making during the li fe cycle of algorithmic systems. These results may appeal mostly to management and organizational researchers and practitioners as well as other social scientists, but they might be challenged by ICT people working on quality appraisal and qualitative design. Divided into a concept building and an empirical part, my own future work will consist of multiple case studies aiming to identify the causes and mitigate the effects of algorithmic opacity in various organizational contexts.
Locomotion automation is a very challenging and complex problem to solve. Besidesthe obvious navigation problems, there are also problems regarding the environmentin which navigation has to be performed. Terrains with...
详细信息
Locomotion automation is a very challenging and complex problem to solve. Besidesthe obvious navigation problems, there are also problems regarding the environmentin which navigation has to be performed. Terrains with obstacles such as rocks, stepsor high inclinations, among others, pose serious difficulties to normal wheeled *** flexibility of legged locomotion is ideal for these types of terrains but thisalternate form of locomotion brings with it its own challenges to be solved, causedby the high number of degrees of freedom inherent to *** problem is usually computationally intensive, so an alternative, using simpleand hardware amenable bio-inspired systems, was studied. The goal of this thesiswas to investigate if using a biologically inspired learning algorithm, integrated in afully biologically inspired system, can improve its performance on irregular terrainby adapting its gait to deal with obstacles in its *** first, two different versions of a learning algorithm based on unsupervised reinforcementlearning were developed and evaluated. These systems worked by correlatingdifferent events and using them to adjust the behaviour of the system so that itpredicts difficult situations and adapts to them beforehand. The difference betweenthese versions was the implementation of a mechanism that allowed for some correlationsto be forgotten and suppressed by stronger ***, a depth from motion system was tested with unsatisfactory results. Thesource of the problems are analysed and discussed. An alternative system based onstereo vision was implemented, together with an obstacle detection system based onneuron and synaptic models. It is shown that this system is able to detect obstaclesin the path of the *** the individual systems were completed, they were integrated together and thesystem performance was evaluated in a series of 3D simulations using various *** simulations allowed to conclude that both learning systems wer
As fishlike underwater vehicle has inspiring maneuverability compared with normal underwater vehicles thrusted by screw propeller, it has much extensive application foreground in cooperative and control for multiple u...
详细信息
ISBN:
(纸本)1424405289
As fishlike underwater vehicle has inspiring maneuverability compared with normal underwater vehicles thrusted by screw propeller, it has much extensive application foreground in cooperative and control for multiple underwater robot. In this paper, an experimental platform of multi mini robofish is developed, and performance test of the robofish is carried out. Rise on this, robofish positioning control was studied. Due to the undulation of water environment and interfere among multi robofish themselves, the swimming of robofish shows highly non-linear and precise positioning control is very difficult. This paper propose a learning algorithm of robofish, and through constructing robofish behavior database, the precise positioning control of robofish is obtained. This study make foundation for multi robofish cooperative control.
The design of neural network architectures is carried out using methods that optimize a particular objective function, in which a point that minimizes the function is sought. In reported works, they only focused on so...
详细信息
The design of neural network architectures is carried out using methods that optimize a particular objective function, in which a point that minimizes the function is sought. In reported works, they only focused on software simulations or commercial complementary metal-oxide-semiconductor (CMOS), neither of which guarantees the quality of the solution. In this work, we designed a hardware architecture using individual neurons as building blocks based on the optimization of n-dimensional objective functions, such as obtaining the bias and synaptic weight parameters of an artificial neural network (ANN) model using the gradient descent method. The ANN-based architecture has a 5-3-1 configuration and is implemented on a 1.2 mu m technology integrated circuit, with a total power consumption of 46.08 mW, using nine neurons and 36 CMOS operational amplifiers (op-amps). We show the results obtained from the application of integrated circuits for ANNs simulated in PSpice applied to the classification of digital data, demonstrating that the optimization method successfully obtains the synaptic weights and bias values generated by the learning algorithm (Steepest-Descent), for the design of the neural architecture.
This paper presents a minimum variance predictive controller (MVPC) using a modified Neural network(MNN) in order to learn the characteristics of a dynamic system. The MVPC can adapt parameters' variation and unce...
详细信息
ISBN:
(纸本)0780342534
This paper presents a minimum variance predictive controller (MVPC) using a modified Neural network(MNN) in order to learn the characteristics of a dynamic system. The MVPC can adapt parameters' variation and uncertainty in the controlled plant through the on-line learning. The learning algorithm is considerably faster because of the introduction of recursive least squares(RLS) algorithm. Simulation results have shown that the proposed approach is effective for adaptive control of nonlinear systems.
暂无评论