Rosa I. Arriaga

An algorithmic theory of learning: robust concepts and random projection

We study the phenomenon of cognitive learning from an algorithmic standpoint. How does the brain effectively learn concepts from a small number of examples despite the fact that each example contains a huge amount of information? We provide a novel algorithmic analysis via a model of robust concept learning (closely related to “margin classifiers”), and show that a relatively small number of examples are sufficient to learn rich concept classes. The new algorithms have several advantages—they are faster, conceptually simpler, and resistant to low levels of noise. For example, a robust half-space can be learned in linear time using only a constant number of training examples, regardless of the number of attributes. A general (algorithmic) consequence of the model, that “more robust concepts are easier to learn”, is supported by a multitude of psychological studies.

Number Discrimination in 10-Month-Old Infants.

Two experiments investigated developmental changes in large number discrimination with visual-spatial arrays. Previous studies found that 6-month-old infants were able to discriminate arrays that differ by a ratio of 1:2 but not 2:3. We found that by 10 months, infants were able to reliably discriminate 8 from 12 elements (2:3) but not 8 from 10 elements (4:5). Thus, number discrimination improves in precision during the first year, and these findings converge with studies using auditory stimuli.

Grow and know: understanding record-keeping needs for tracking the development of young children

From birth through age five, children undergo rapid development and learn skills that will influence them their entire lives. Regular visits to the pediatrician and detailed record-keeping can ensure that children are progressing and can identify early warning signs of developmental delay or disability. However, new parents are often overwhelmed with new responsibilities, and we believe there is an opportunity for computing technology to assist in this process. In this paper, we present a qualitative study aimed at uncovering some specific needs for record-keeping and analysis for new parents and their network of caregivers. Through interviews and focus groups, we have confirmed assumptions about the rationales parents have and the functions required for using technology for record-keeping. We also identify new themes, potential prototypes, and design guidelines for this domain.

Baby steps: evaluation of a system to support record-keeping for parents of young children

Parents of young children often want to keep a variety of records on their childrens early years, for the purposes of preservation of memories or at the request of their pediatrician. However, time constraints, motivation, and forgetfulness may hinder their ability to keep consistent records. We developed a system, Baby Steps, which is designed to improve the record-keeping process. In this paper, we present the results of a 3-month deployment study of this technology with 8 families and their pediatricians. The study showed that when compared to a control condition, experimental design features of Baby Steps encouraged parents to more frequently collect and review records, provided higher confidence in reporting, and improved parent-pediatrician communication.

Startle potentiation by threat of aversive stimuli and darkness in adolescents: a multi-site study

In recent years, the startle reflex has become an exciting new tool to investigate affective responses to aversive stimuli in humans. The popularity of this methodology is largely based on the substantial amount of animal research available on this topic. Several procedures have been developed to examine startle potentiation in humans, but most studies have been carried out in adults and may not be appropriate for children or adolescents. The present study is a multi-site project (Yale University, Harvard University, and the University of Minnesota) investigating two new procedures to examine the potentiation of startle in adolescents. The subjects were 50 male and female aged 13-17 years old. One procedure examined fear-potentiated startle to the threat of an unpleasant airblast directed to the larynx. The second examined the facilitation of startle in darkness. Potentiation was found using each procedure and the degree of potentiation was similar across laboratories. These results suggest that both the threat of an airblast and darkness can reliably be used to examine startle potentiation in young subjects.

The effects of a community-based literacy program on young children's language and conceptual development

Effects of a community-based literacy program on 1−, 2−, and 3-year-old childrens language and conceptual development were assessed. University students were trained to teach Head Start parents effective methods for reading to their children. Families were randomly assigned to receive 18, 3, or 0 instructional visits. Results indicated that parents in the 18-instructional-visit program increased their participation in appropriate literacy behaviors such as reading to their children, teaching concepts to their children, and using the library, more than parents in the 0-instructional-visit groups. Children in the 18-instructional-visit program showed greater gains in language and conceptual development than children in the 0-instructional-visit group. Few differences were found between children in the 3-visti and 0-instructional-visit groups. Thus, only a high-intensity community-based intervention designed to train parents was effective in increasing emergent literacy in low-income ethnic children.

Designing a social network to support the independence of young adults with autism

Independence is key to a successful transition to adulthood for individuals with autism. Social support is a crucial factor for achieving adaptive self-help life skills. In this paper we describe the results of a formative design exercise with young adults with autism and their caregivers to uncover opportunities for social networks to promote independence and facilitate coordination. We propose the concept of SocialMirror, a device connected to an online social network that allows the young adult to seek advice from a trusted and responsive network of family, friends and professionals. Focus group discussions reveal the potential for SocialMirror to increase motivation to learn everyday life skills among young adults with autism and to foster collaboration among a distributed care network. We present design considerations to leverage a small trusted network that balances quick response with safeguards for privacy and security of young adults with autism.

Exploiting social partners in robot learning

Social learning in robotics has largely focused on imitation learning. Here we take a broader view and are interested in the multifaceted ways that a social partner can influence the learning process. We implement four social learning mechanisms on a robot: stimulus enhancement, emulation, mimicking, and imitation, and illustrate the computational benefits of each. In particular, we illustrate that some strategies are about directing the attention of the learner to objects and others are about actions. Taken together these strategies form a rich repertoire allowing social learners to use a social partner to greatly impact their learning process. We demonstrate these results in simulation and with physical robot ‘playmates’.

Using SMS to provide continuous assessment and improve health outcomes for children with asthma

Effective communication between health professionals and patients positively influences chronic health management, as does increased patient awareness of their symptoms and general knowledge of the condition. In this study, we leverage the use of mobile phones by pediatric patients and report results from a four-month randomized controlled trial (RCT). We examined: 1) how a SMS system impacted the health outcomes of asthmatic children; and 2) how physicians used a Web service showing the data gathered from the SMS system. Our results show that 1) the simple act of communicating knowledge and symptom awareness information via SMS leads to improved pulmonary function for pediatric patients; and 2) physicians would use the data sent from the SMS system to monitor their patients asthma management status.

Socially computed scripts to support social problem solving skills

The social world that most of us navigate effortlessly can prove to be a perplexing and disconcerting place for individuals with autism. Interactive tools to teach social skills that are personalized to the individuals needs show promise, but it is challenging to author them. We describe the design, development, and preliminary evaluation of an approach to using human computation that enables the creation of models of complex and interesting social scenarios, possible obstacles that may arise in those scenarios, and potential solutions to those obstacles. Our preliminary evaluation of the models confirms that these models have the potential to help an author create a social skills instructional module.