User Experience and Interaction Design
Dr. Bo Fu, Assistant Professor, CSULB
Traditional approaches to designing and developing computing systems typically emphasize rigorous testing of correct system behavior, e.g. verifying that a set of system functionalities work to their specification, while interaction design is often ignored, e.g. a lack of usability and user experience testing to evaluate how well users are interacting with a system. In reality, it is often a usability or user experience issue that actually prevents a user from successfully completing a task rather than a critical system error. This workshop introduces students to designing and running experiments on user interfaces of their choice for the purpose of identifying and resolving usability issues in user-centered design and development.
Analyzing acoustic telemetry data to track sharks
Dr. Alvaro Monge, Professor, CSULB
The SharkLab at CSULB uses acoustic telemetry to study the movement of animals along the coastal waters of Southern California. They study several species, including, of course, sharks. In this project, students will visit the Shark Lab, interact with researchers in the lab, learn about acoustic telemetry and the data that has been collected using these techniques, and the research questions they’re trying to answer about the animals they’re tracking. Students will be given a data set and be asked to propose their own research questions and ways to analyze and visualize the data.
Resilient Smart Gardens
Dr. Birgit Penzenstadler, Assistant Professor, CSULB
Despite rising interest in being able to grow food ourselves for educational as well as sustainability purposes, in drought-prone Southern California, it is challenging to grow herbs and vegetables. In this project, students learn how to build a mobile app that will help design and monitor a garden according to permaculture principles. To expand on that scope, students can tie the system in with automated watering using an Arduino board fueled by solar energy.
Deep Learning for 3D Neuron Segmentation of Microscopy Images
Dr. Wenlu Zhang, Assistant Professor, CSULB
Deep learning is recently considered as one of the most breakthrough technologies in computer science. Deep learning models have been successfully applied to a variety of real-world applications such as natural image classification, generation and detection. However, in the biomedical field, deep learning is still very challenging due to the multi-modality of patterns and the incomplete medical or biological data. This workshop will introduce students to implementing and designing deep learning models in 3-dimensional (3D) microscopy images. A central challenge in neuroscience is to identify the 3D morphology of neurons from microscopy images. Students will first understand the basic concepts and techniques in deep learning, such as convolutional neural network (CNN), recurrent neural network (RNN) and generative adversarial network (GAN), etc. They will also learn how to develop 3D encoder-decoder neural network architectures to train and predict using large microscopy images in an end-to-end manner.
Teaching Computers to Reason about Programs
Dr. Sorin Lerner, Professor, UCSD
For decades humans have been writing programs and reasoning about their behavior, both on a formal and informal level. With the development of modern "proof assistants", we can now write down that reasoning at a level where it can be automatically checked by a computer. But producing such proofs is still the task of an expert programmer or formal methods expert. In this project we'll explore building a system that attempts to progress proofs automatically, automating much of the tedious proving work by finding novel patterns in large proof bodies.
Bio-Inspired Pattern Formation With a Set of Robots
Dr. Oscar Morales-Ponce, Assistant Professor, CSULB
Self-configurable organisms are fundamental to create complex structures. Consider for example unicellular organisms. Each cell reacts according to the substance that the neighbors produce. This process continues until the cells produce a stable multicellular organism. In other words, a global problem is solved only with local decisions. Previous research such as Cellular Automata has modeled the problem using a dynamic grid where the state of the current step is a function of the state of the previous step. In this project, we model the problem using a set of low-power robots called Kilobots that represent simple organisms. Initially, robots can be in any state. For simplicity, the state is a set of k integers. Each state may represent a simple behavior such as motion, stationary, red, green, etc. In every step, robots communicate their own state to the nearest neighbors. Each robot then computes the new state according to the state of its neighbors that it receives in the previous step. This project has two main objectives. The first is to build a robotic framework to perform extensive experiments. The second is to obtain a deterministic function that guarantees that the robots form a stable complex pattern after a finite number of steps.
Where is Hollywood?: Artificial Intelligence approach
Dr. Ju Cheol Moon, Assistant Professor, CSULB
Friends from my hometown in South Korea are visiting Los Angeles for the first time and they want to go to Hollywood today. We know about the famous intersection of Hollywood Boulevard and Vine Street, but is that where Hollywood really is? Do these streets represent the Hollywood we’ve come to know in movies and on social media, or is there a better map that marks the region we’re looking for? During this workshop, through the analysis of Twitter data, I will demonstrate how to delineate specific regions by using artificial intelligence methodologies and natural language processing techniques. Changes in regions based on the sentiment or time detected in tweets are also taken into consideration in the process of delineation. Participants of the project can expect to learn artificial intelligence methodologies (including clustering algorithms), natural language processing techniques (including sentiment analysis), evaluation criteria of the designed algorithm, and usage of the Twitter API and the Google API.