ApproxNet: Content and Contention-Aware Video Object Classification System for Embedded Clients

Videos take a lot of time to transport over the network, hence running analytics on the live video on embedded or mobile devices has become an important system driver. Considering such devices, e.g., surveillance cameras or AR/VR gadgets, are resource constrained, although there has been significant work in creating lightweight deep neural networks (DNNs) for such clients, none of these can adapt to changing runtime conditions, e.g., changes in resource availability on the device, the content characteristics, or requirements from the user. In this article, we introduce ApproxNet, a video object classification system for embedded or mobile clients. It enables novel dynamic approximation techniques to achieve desired inference latency and accuracy trade-off under changing runtime conditions. It achieves this by enabling two approximation knobs within a single DNN model rather than creating and maintaining an ensemble of models, e.g., MCDNN [MobiSys-16]. We show that ApproxNet can adapt seamlessly at runtime to these changes, provides low and stable latency for the image and video frame classification problems, and shows the improvement in accuracy and latency over ResNet [CVPR-16], MCDNN [MobiSys-16], MobileNets [Google-17], NestDNN [MobiCom-18], and MSDNet [ICLR-18].

Simulations as a teaching strategy in Social Work training

This paper is about the simulations' systematization and evaluation, implemented over 13 years, as part of the training of Social Work students at the University of Costa Rica. As a product of the systematization, the simulations' theoretical foundation is detailed, integrated by student-centered approach, situated learning, and cultural-historical theory. Besides, the didactic process had two moments: first, the strategy preparation and implementation, and second the debriefing with all the participants. The evaluation was implemented in 2017, 2018, and 2019. A questionnaire was developed and completed by 60 students in these three years. The students consider the simulations as a more dynamic approximation to the knowledge provided by the course, through the practical approach to situations they will face in their professional practice. The students acquired several learnings with this training, on a theoretical, methodological, and technical level, as well as the development of skills for the attention of crises.

ON DIALOGUES AND EXISTENCES: A POSSIBLE CONTRIBUTION OF ANTHROPOLOGY TO AGROECOLOGY

We intend to problematize the agroecological “dialogue of knowledge” and survey the conditions of anthropology for such problems. The fundamental argument is that, despite the proposal of non-hierarchical construction of agroecological knowledge (between “scientific” and “non-scientific” knowledge), its epistemological basis does not comprises the complex interactions produced among the agents in question, ambivalently contributing to a hierarchical and asymmetric relationship of power. We suggest that: the idea that agroecology promotes harmonious relations between scientists and farmers simplifies interactions tensioned by the dynamic approximation/estrangement; the conception of culture and cultural diversity implicit in the dialogue of knowledge is based on the dissociation between “thought” and “action”; a “political dynamics of alterity” would better define the agroecological relational scheme; the anthropological conception of translation would contribute to the promotion of a “double extensionality,” not based on the capacity to teach, but rather on the capacity to learn.