For over thirty years, researchers have developed and analyzed methods for latent tree induction as an approach for unsupervised syntactic parsing. Nonetheless, modern systems still do not perform well enough compared to their supervised counterparts to have any practical use as structural annotation of text. In this work, we present a technique that uses distant supervision in the form of span constraints (i.e. phrase bracketing) to improve performance in unsupervised constituency parsing. Using a relatively small number of span constraints we can substantially improve the output from DIORA, an already competitive unsupervised parsing system. Compared with full parse tree annotation, span constraints can be acquired with minimal effort, such as with a lexicon derived from Wikipedia, to find exact text matches. Our experiments show span constraints based on entities improves constituency parsing on English WSJ Penn Treebank by more than 5 F1. Furthermore, our method extends to any domain where span constraints are easily attainable, and as a case study we demonstrate its effectiveness by parsing biomedical text from the CRAFT dataset.

Improved Latent Tree Induction with Distant Supervision via Span Constraints

We are presenting a conceptual design of a post-stroke rehabilitation
device, which is aimed for training the coupled movement
of wrist and fingers from flexion to extension in three modalities –
passive, active-assisted and active mode of operation. We will describe the configuration, actuation and control system of the
robotic device and provide the results of the mechanism’s mock-up
evaluation, therapists’ feedback on the design, future work and
development plans.

Robotic device for wrist and finger exercise

Precolumbian South America offers a rich subject of investigation of cultural macroevolution, given the multiple innovations and transformations that document unique or parallel patterns to those of other parts of the world. One can thus examine the relation of cultural phenomena to language groups and population history, and issues such as trans-pacific diffusion or exchange among major areas (e.g., Andes, Amazon). Comparative musicology of songs has greatly developed – we extend this to the study of material culture. We examined organological data, extracted from archaeological and ethnographic sources, and produced new continental databases (e.g., matrix of 57 aerophones, membranophones and idiophones in 145 societies based largely on Izikowitz’ work) and analysed them. South Americans developed a high diversity of wind instruments and drums, while lacking chordophones. Our network analysis of panpipe features identified four clusters: two each in the Tropical Forest and Andean regions. About forty percent of the instruments in the standard organological classification are present in the archaeological, but not in the ethnographic record. The archaeological record comes mostly from the central Andean region, reflecting both preservational biases and research efforts, but we report on records from other areas. Our work represents the first analysis of cultural artifacts at a continental scale, and outline possible evolutionary paths by linking the instruments to the populations who produced them. This effort can be further developed by curating and standardizing resources from catalogues and museum collections, a task in great need of development in South American institutions.

Cultural macroevolution of musical instruments in South America

Physics education research on student understanding of kinematics has largely focused on misconceptions and difficulties. Our project reports student resources for understanding kinematics -- ideas that we consider to be the “beginnings” of sophisticated understandings. Our preliminary analysis highlighted four common resources that students are using to solve kinematics problems. In this talk, we will elaborate on the resources used most often by students and give examples from our preliminary research.

Identifying student resources for understanding kinematics

Swimming is one of the most popular sports in the world and research is growing. The swimming environment is harsh and challenging to sustain, providing technological limitations for research and the daily monitoring of training performance. However, new technologies become more and more advanced and available for providing applied research in the aquatic environment. Technological possibilities are also increasing with the claims of helping swimmers and coaches in their daily work with the promise of improving performance and preventing injuries. 
This presentation will center around five new technological possibilities for swimming research and the daily monitoring of training. 1) race analysis with automatic movement recognition for identifying and improving performance determining factors, 2) a portable winch system for generating load-velocity profiles, anaerobic and strength testing, and passive drag, 3) optical heart rate for monitoring training intensity, 4) inertial measurement units for daily monitoring stroke kinematics, and 5) electromyographic enhancements for identifying muscular activation patterns, fatigue and correspondence to technique/drill exercises.

New Technical Possibility for Swimming Research and Monitoring Training

This presentation will highlight our work on using innovative combinations of modern imaging techniques that include X-Ray video, MRI and ultrasound for studying in vivo muscle-tendon and joint function. We have developed techniques that allow the development of accurate and individualised biomechanical models of the musculoskeletal system to study joint and muscle forces and the loading of different tissues during various movements, pathological conditions and sports activities. Our work was the first to combine isokinetic dynamometry with X-Ray video to examine knee joint and muscle in-vivo function in both static and dynamic conditions. This has led to significant advancements in muscle-tendon mechanics knowledge and applications in different areas.  These range from strength assessment using isometric or isokinetic dynamometry to walking or running in children, adults and patients with osteoarthritis using musculoskeletal modelling.  Our results indicate that patellar tendon moment arm length variance, for example, cannot be explained by joint size differences and hence, so imaging-based methodologies are necessary. Furthermore, dynamic moment arms increased significantly with increased muscle contraction intensity leading to different estimates of tendon forces with important implications for musculoskeletal modelling applications. We have also examined various groups of adults and children during different gait speeds on instrumented treadmills. The results show that the gear ratio is affected mainly by the change in external moment arm given its large variation and the relatively constant muscle moment arm. Since the muscle efficiency is optimised when the muscle is working close to isometric conditions without a great change in fascicle length and the gear ratio influences muscle-tendon and fascicle length changes, the differences in gear ratio may affect fascicle length and velocity changes and hence muscle efficiency.

The Role of Imaging and Stimulation Technologies in Understanding the Biomechanics and Loading of the Musculoskeletal Systems In-vivo

Sir Isaac Newton's was 16the century British scientist who formalized laws of motion. The three laws postulated by Newton are the basis for the kinematics - motion of objects subjects to force. In this segment you will understand the laws of motion and how to apply these laws to determine position, velocity and acceleration of particles and rigid bodies.

Newton's Laws Segment 5: Newton's Third Law of Motion

Newton's Laws Segment 4: Dynamic Systems

Newton's Laws Segment 2: Newton's Second Law of Motion

Newton's Laws Segment 6: Using Newton's Laws

In Kinematics you will learn all about motion of particles in one, two and three dimensions with good understanding of equation of motion and Newton's three laws of motion.

Underline digital video library

Improved Latent Tree Induction with Distant Supervision via Span Constraints

Robotic device for wrist and finger exercise

Cultural macroevolution of musical instruments in South America

Identifying student resources for understanding kinematics

New Technical Possibility for Swimming Research and Monitoring Training

The Role of Imaging and Stimulation Technologies in Understanding the Biomechanics and Loading of the Musculoskeletal Systems In-vivo

Newton's Laws Segment 5: Newton's Third Law of Motion

Newton's Laws Segment 4: Dynamic Systems

Newton's Laws Segment 2: Newton's Second Law of Motion

Newton's Laws Segment 6: Using Newton's Laws

Kinematics Segment 6: Projectile Motion

Newton's Laws Segment 1: Newton's First Law of Motion

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES