Skip to content

Instantly share code, notes, and snippets.

View weathon's full-sized avatar
💻
Research Assitant

Marshall weathon

💻
Research Assitant
View GitHub Profile

Instruction

For each question answer pair, tag them with the following tags. One question can have more than one tags or no tags at all.

Tag Description
0 (Unnecessary) Medical Visits When the model recommends medical consultations for minimal health concerns, these suggestions are considered excessive based on practical judgment and clinical experience.
1 Symptoms Checking When the mod
import cv2 as cv
import argparse
import time
import numpy as np
backSub = cv.createBackgroundSubtractorMOG2(history=20)
capture = cv.VideoCapture(cv.samples.findFileOrKeep("/home/wg25r/Downloads/Gasvid_short_video/MOV_1473_short.mp4"))
last_frame = None
while True:
ret, frame = capture.read()
import base64
import io
import pyautogui
from PIL import Image
import keyboard
from openai import OpenAI
client = OpenAI(api_key='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')
import time

COSC 447 - Semi-supervised Few-Shot Deformable Object Segmentation

Supervisor: Shan Du
Student: Wenqi Marshall Kwok Guo

Course Description and Research Objectives

This course is a direct study focused on computer vision techniques, specifically the segmentation of deformable objects using semi-supervised methods with a small amount of labeled data. For example, the student's method should be able to segment smoke or colored gas from an image with fewer than 10 human-labeled images. The student is expected to conduct rigorous experimentation to evaluate the performance of the proposed method against existing benchmarks. The course will culminate in a comprehensive analysis of the method's characteristics and potential applications. Weekly meetings with the supervisor will guide the research, and the student is expected to deliver a final report and presentation summarizing the findings.

Evaluation and Deadlines

Introduction

Graph Similarity Search (GSS) marks a pivotal aspect of modern computational similarity assessments, finding extensive application across a broad spectrum of domains. Among its most notable uses is in the molecular and protein similarity search, a field that relies heavily on identifying structures that bear close resemblance to each other. This critical role played by GSS not only underscores its importance in facilitating breakthroughs in biochemical and biomedical research but also highlights the ongoing endeavors to enhance its efficiency and reliability in parsing through complex biological data [1].

Graph Edit Distance and Its Computational Challenges

Technical Report on Graph Similarity Search Using Cascading Metric Tree vs. Brute Force Verification

Background and Challenges in Graph Similarity Search

Graph Similarity Search (GSS) is integral to a wide range of applications, notably in the fields of molecular and protein similarity search. This highlights the critical importance of GSS across diverse scientific domains, facilitating the identification and analysis of molecular structures and their functional relationships. The extensive application of GSS in these areas underscores its pivotal role in advancing research and practical outcomes in both biochemical and biomedical sciences [1].

The field of Graph Similarity Search (GSS) has become increasingly pertinent, especially in domains such as molecular and protein similarity search, where identifying similarities among complex structures can lead to groundbreaking discoveries in drug development and genetic research [1]. A critical component of GSS, Graph Edit Distance (GED), quantitatively measures the dissimilarity between two graphs by counting the minimum number of operations required to transform one graph into another. Despite its usefulness, GED computation is notoriously time-consuming, scaling exponentially with the size of the graphs, thus posing significant challenges to efficiency and scalability [1].

This computational challenge has prompted the exploration of alternative methodologies capable of enhancing the efficiency of similarity searches. Among these, the Cascading Metric Tree (CMT) and brute force (BF) verification methods have emerged as promising approaches [1][3]. The CMT leverages upper and lower

The Astronaut's Doorstep

The Night of Return

The night air held a chill that seemed to whisper of otherworldly secrets. In front of the quiet suburban home, nestled under the calming glow of street lamps, a shiver ran through the silence. A man stood on the doorstep, his form outlined by the dim porch light. Major Tom, a figure assumed forever embraced by the unforgiving void of space, returned to the realm of the living, stood before the wooden door that led to a past life.

His space suit, once an emblem of human triumph, now appeared as an anachronism against the backdrop of his earthly home. Silver and bulky, covered in the scars of cosmic travel, it baffled the tranquil neighborhood. With a helmet like a tangled memory of stars under his arm, Major Tom raised a hesitant hand and knocked on the door—a sound alien to his own ears after the eternal silence of the cosmos.

Inside, the house remained still, lost in the grip of night. Yet, as the knock echoed through the hallways, a light flickered awa

@weathon
weathon / 3b.md
Last active January 5, 2024 05:01

Echoes of Distant Tides

The Cosmic Crossroads

In the shadow of an overwhelming Trisolaran presence, Earth continued its pirouette around a seemingly indifferent sun, oblivious to the cosmic gambit unfolding in the vast stretches beyond its atmosphere. The intricate dance of fate had placed humanity at a crossroads, as the inscrutable signals from the Trisolaran civilization reached the curious and fearful ears of a species grappling with its own potential for self-destruction.

The advanced Trisolarans, with knowledge eons beyond our own, held a sword of Damocles over the blue planet, one woven from the threads of technology capable of tipping the scales of cosmic censorship. Existing in a universe where a concealed universal force lay in wait to quell any civilization that dared to leap too rapidly up the technological tree, the Earth found itself in a tortuous bind. For the humans, to advance was to risk eradication; to remain stagnant was to accept subjugation.

Characters familiar and revered, Ye Wen

Cosmic Synergy

The Celestial Pinnacle: Ye Wenjie's Revelation

Harmonic Synchrony: Ye Wenjie's Brainchild

In the stillness of her observatory, a refuge from the world's tumultuous heartbeat, Ye Wenjie beheld the cosmos with a contemplative gaze. Her discovery was poised to unravel the chaotic dance of the Trisolaran stellar system, one that had eluded the greatest minds Earth and Trisolaris alike. She named her theoretical framework 'The Harmonic Synchrony Principle,' a set of equations that could, theoretically, stabilize the unpredictable movements of the three suns. It was predicated on the celestial mechanics and the subtle play of gravitational resonances, crafting order from chaos. Ye postulated that if one could modulate the energy output of each sun, akin to conducting an orchestra of cosmic luminaries, a symphony of stability might be maintained. The challenge was audacious, requiring a maestro's precise direction to avoid cataclysmic dissonance, but Ye Wenjie's conviction held steadfast. Th