[note: this site is currently under development]

Hi, my name is Mason Wang.

I'm a junior at Saratoga High School. I'm an experienced full-stack developer and AI/ML enthusiast. I'm currently focused on ML research.

I love applying my skills at hackathons and have won top prizes at every single one I've attended, including prestigious collegiate ones such as UCLA's (1st place overall) and MIT's (3rd place overall). At school, I strive to help other students discover the power of computer science as the co-president of both the CS and AI clubs. I'm also very involved in schoolwide leadership activities and currently serve as the ASB secretary.


Research

Sparse Fusion for Multimodal Transformers

Yi Ding*1, Alex Rich*1, Mason Wang2, Noah Stier1, Matthew Turk3, Pradeep Sen1, Tobias Höllerer1
1UC Santa Barbara, 2Saratoga High School, 3Toyota Technological Institute at Chicago

We present Sparse Fusion Transformers (SFT), a novel multimodal fusion method for transformers that performs comparably to existing state-of-the-art methods while demonstrating up to a six-fold reduction in computation and memory requirements.

Currently under review.

View paper

Interpolating Latent Representations to Increase Label Noise Tolerance

Mason Wang1, Yi Ding2
1Saratoga High School, 2UC Santa Barbara

We propose the use of a modality-agnostic label noise tolerant technique involving training on interpolated latent representations along with their labels, encouraging neural networks to predict less confidently on these in-between examples. Our technique reduces the memorization of corrupt labels, which can alleviate the issue of noisy labels.

View paper

CLEFNET: Recurrent Autoencoders with Dynamic Time Warping for Near-Lossless Music Compression and Minimal-Latency Transmission

Vignav Ramesh*1, Mason Wang*1
1Saratoga High School
(authors contributed equally)

We propose an architecture (Latent Space) for a platform with end-to-end, near-lossless audio transmission tailored specifically to online joint music production. We discuss the usage of a recurrent autoencoder (ClefNet), propose a new evaluation metric (MSE-DTW), and detail the logistics of a live system implementation.

View Paper

Projects

Archiscape

1st Place Overall @ LAHacks 2020

Archiscape is a PWA aimed at revolutionizing the fields of smart infrastructure, urban planning, and home design. Won 1st Place Overall at LAHacks 2020 (1,100+ participants), SoCal's largest hackathon and one of the world's largest collegiate hackathons. Featured in CBS News, The UCLA Newsroom, Build for COVID-19, Scommerce, and The Saratoga Falcon.

Learn More, Github, Video Demo

Latent Space

3rd Place Overall @ HackMIT

Latent Space is a decentralized video calling web app that utilizes recurrent neural autoencoders with embeddings specifically for music to transfer encoded (compressed) audio files with minimal latency.

Won 3rd Place Overall at HackMIT 2020 (1,100+ competitors), MIT's largest undergraduate hackathon and one of the world's largest collegiate hackathons. Featured on Medium.

Video Demo Github, Website, Finalist Presentation

Centivize

3rd Place Overall @ COVIDathon 2020

Centivize is a blockchain/AI based PWA that allows users with conditions less critical than COVID-19 to obtain diagnoses based on their medical histories and requests that volunteers are incentivized to fulfill via cryptocurrency.

Won 3rd Place Overall at COVIDathon 2020, the world's largest decentralized AI workplace hackathon. Worked with the Platform Team at SingularityNET to publish Centivize's summary and similarity algorithms on SNET's dApp: available here. Centivize will appear on the OceanProtocol compute to data platform in the near future.

Github, Devpost, Video Demo, Ocean Protocol Proposal

Other

Other hackathon projects on Devpost (100% hackathon win rate). Placed 1st Place @ IgniteCS Programming Expo in 2019 (200+ competitors).


Education

Saratoga High School

August 2019 - June 2023

GPA: 4.42/4.00

Foothill College

Concurrently Enrolled

GPA: 4.00

West Valley College

Concurrently Enrolled

GPA: 4.00

UC Santa Barbara

Research Mentorship Program

Certificates

Contact

Contact form + links.