Useful links
- website
- https://ideas.repec.org/s/cen/wpaper.html
- https://www.census.gov/about/adrm/fsrdc.html
Author: samuel.cheng@ou.edu
See link here. Due Oct 31.
The Norman chapter of P.E.O. (a women’s philanthropic organization) is looking for an outstanding female PhD candidate. Their mission is to support women in the attainment of their educational goals. The P.E.O. Scholars Award (PSA) a $20k grant for PhD students within two years of completion of their degree, with a well-defined research project and within at least a year to completion: https://www.
Tulsa students now can directly apply for travel grants in Norman.
Three Faculty Positions in Data Science: Human-Computer Teaming and Interactive Decision Making; Artificial Intelligence Architectures; and Trustable Artificial Intelligence at the University of Oklahoma, Norman Campus
Positions Available: As part of a multiyear effort to grow world-class data science and data enabled research across The University of Oklahoma (OU), the Gallogly College of Engineering (GCoE), Department of Electrical and Computer Engineering and/or Department of Computer Science, in partnership with the Dodge Family College of Arts and Sciences (CAS), welcomes applications for a cluster of three (3) faculty positions from candidates whose experiences and interests have prepared them to be an integral contributor engaged in scientific discovery, developing talent, solving global challenges, and serving our society. This year we are focusing on data science foundational and enabling technologies. In subsequent years, we’ll be hiring additional data science and data-enabled research faculty.
The University, as part of its Lead On, University strategic plan has committed to creating world class capabilities in data science, artificial intelligence (AI), machine learning (ML), and data enabled research. In July 2020, the University established the Data Institute for Societal Challenges (DISC) to grow convergent data-enabled research to solve global challenges. DISC currently has over 130 faculty members across OU campuses, nine communities of practice, seed funding programs, and an extended network of approximately 300 data scientists and data enabled researchers across many disciplines (https://www.ou.edu/disc).
Three positions:
1) Professor or Associate Professor in Human-Computer Teaming and Interactive Decision Making: Humans and computers have complementary knowledge and skillsets. To solve challenging problems, we need to team this expertise together for effectiveness, reliability, efficiency, and adoption of many data-driven solutions. This area is cross
disciplinary, and we seek a senior faculty member with expertise in one or more of human computer teaming, visualization, visual analytics, human-machine interaction, decision theory, HCI, human factors and industrial engineering, or cognitive psychology. This faculty member will be a vital core team member in data science and data-driven decision making with a home department in ECE and possible joint appoint in ISE, Computer Science, Psychology, and/or Political Science.
Applications should be submitted online via Interfolio at
http://apply.interfolio.com/112374. Inquires can be addressed to Professor David Ebert, chair of the search committee at ebert@ou.edu.
2) Assistant Professor in AI Architectures: We seek to recruit a transdisciplinary faculty member with expertise in one or more of the following areas: scalable, high-performance software and hardware architectures for AI and advanced analytics, advanced and domain-tailored data science, AI (trustable, science-based, and human-guided), and human-computer teaming. Specific areas of interest include probabilistic, neuromorphic, and novel architectures, software pipelines and operating system architectures to support high-performance analytics, and enable real-time trustable AI and decision-making. Since traditional computing architectures are still based on solving problems from the 20th century, new computing hardware and software architectures are needed to optimize computing for AI and machine learning and many new approaches to science and engineering. This faculty member will grow and complement work in computer engineering, computer science and the new OU quantum center (CQRT) with a home department in ECE and possible joint appointments where appropriate.
Applications should be submitted online via Interfolio at
http://apply.interfolio.com/112359. Inquires can be addressed to Professor David Ebert, chair of the search committee at ebert@ou.edu.
3) Assistant Professor in Trustable AI. We are seeking an Assistant Professor in Trustable AI. Human-guided, science-based, explainable AI (xAI) are key areas to ensure AI is understandable, reliable, and robust for real-world applications. This faculty member will grow our expertise in one of the most rapidly developing and vital fields of data science, with a primary home in ECE and potentially joint appointments in CS, Psychology, and ISE. We seek a faculty member with expertise in one or more of science-based AI or machine learning (ML), human-guided AI/ML, explainable AI/ML, and closely related topics. This faculty member will be a vital core team member in data science, AI, and data-driven convergent research solutions to global challenges. This faculty member will provide vital capabilities that will empower research in all four strategic verticals and grow the data science ecosystem on campus to create the critical mass in data science needed for the success of the university’s strategic plan, Lead On, University.
Applications should be submitted online via Interfolio at
http://apply.interfolio.com/112372. Inquires can be addressed to Professor David Ebert, chair of the search committee at ebert@ou.edu.
Gallogly College of Engineering: The mission of the GCOE is to foster creativity, innovation and professionalism through dynamic research, development and learning experiences.
The Gallogly College of Engineering is home to the Data Science and Analytics Institute (https://www.ou.edu/coe/dsai). The DSAI provides undergraduate and graduate certificates, Master’s degrees, and PhD degrees in data science and analytics as well as offering workforce upskilling to industry partners. Faculty members in GCoE and across campus participate in the DSAI.
The University of Oklahoma: OU is a Carnegie-R1 comprehensive public research university known for excellence in teaching, research, and community engagement, serving the educational, cultural, economic and healthcare needs of the state, region, and nation from three campuses: Norman, Health Sciences Center in Oklahoma City and the Schusterman Center in Tulsa. OU enrolls over 30,000 students and has more than 2700 full-time faculty members in 21 colleges.
Norman is a vibrant university town of around 113,000 inhabitants with a growing entertainment and art scene. With outstanding schools, amenities, and a low cost of living, Norman is a perennial contender on “best place to live” rankings.
Visit http://www.ou.edu/flipbook and http://soonerway.ou.edu for more information. Within an easy commute, Oklahoma City features a dynamic economy and outstanding cultural venues adding to the region’s growing appeal.
Qualifications
Successful candidates must have the interest and ability to contribute significantly to the advancement of these fields and develop a nationally recognized program of sponsored research; teach at both the undergraduate and graduate levels; supervise graduate students and postdoctoral fellows. A Ph.D. in computer science, engineering, or related discipline is required.
Application Instructions
Confidential review of applications will begin October 1, 2022. Candidates are invited to submit a
letter of interest, names of three references who will be contacted only upon approval from the applicant, curriculum vitae, and brief (~2-3 pages) statements of interest regarding 1) research, 2) teaching, and 3) service. The research statement should summarize your prior contributions to research and your goals for developing a research program at OU. The teaching statement should summarize past instructional and mentorship experiences, and plans/goals for teaching at OU (including existing and proposed courses) and advising a varied cohort of undergraduate and graduate students. The service statement should describe your vision for internal service to the academic unit, the College and the University, and for external service to our scientific community and other stakeholders. Candidates are requested to submit their applications to the appropriate position listed above and inquiries should be directed to the search committee chairs, also listed above.
Inquiries should be directed to the search committee chair:
Dr. David S. Ebert, Gallogly Chair Professor
School of Electrical and Computer Engineering and School of Computer Science Associate Vice President of Research and Partnerships
Director, Data institute for Societal Challenges
University of Oklahoma
Email: ebert@ou.edu
Equal Employment Opportunity Statement
The University of Oklahoma, in compliance with all applicable federal and state laws and regulations does not discriminate on the basis of race, color, national origin, sex, sexual orientation, genetic information, gender identity, gender expression, age, religion, disability, political beliefs, or status as a veteran in any of its policies, practices, or procedures. This includes, but is not limited to: admissions, employment, financial aid, housing, services in educational programs or activities, or health care services that the University operates or provides.
Diversity Statement
The University of Oklahoma is committed to achieving a diverse, equitable and inclusive university community by recognizing each person’s unique contributions, background, and perspectives. The University of Oklahoma strives to cultivate a sense of belonging and emotional support for all, recognizing that fostering an inclusive environment for all is vital in the pursuit of academic and inclusive excellence in all aspects of our institutional mission.
Mission of the University of Oklahoma
The Mission of the University of Oklahoma is to provide the best possible educational experience for our students through excellence in teaching, research and creative activity, and service to the state and society.
I have a hard time standing manuscriptcentral. It is an absolute piece of crap. And unfortunately, it seems that the entire community is stuck with it. I am serving for AE for TCSVT this year. Like many journal publications, it was stuck with the crappy manuscriptcentral.
Just to be realistic, the hit rate of getting a reviewer is very low this day. I have created another tool to extract emails from hundreds of papers. I will now input close to 100 emails per manuscript to secure enough reviewers. It can take at least an hour to do that as manuscriptcentral do not have an import function. And the antiquated UI even expects an editor to input an email one by one.
So I tried to automate that using selenium. I am not an expert on that, but I saw another project using it and so tried to imitate what he did. It is part of the code below (after log in the page of the respective manuscript). And par is email extracted by my other tools. Basically, a csv file with the first two columns are the emails and names of the reviewers. The code is not very refined, but it did the trick.
from selenium.webdriver.common.by import By
for line in par.split('\n')[8:]:
email=line.split(',')[0]
firstname=line.split(',')[1].strip().split(' ')[0]
lastname=line.split(',')[1].strip().split(' ')[1]
driver.switch_to.window(driver.window_handles[0])
url='javascript:openCreateAccountPopup()'
addReviewer = driver.find_element_by_xpath('//a[@href="'+url+'"]')
addReviewer.click()
driver.switch_to.window(driver.window_handles[1])
# css_firstname = driver.find_element_by_css_selector('input[name="PERSON_FIRSTNAME"]')
css_firstname=driver.find_element(By.NAME, "PERSON_FIRSTNAME")
css_firstname.send_keys(firstname)
css_lastname=driver.find_element(By.NAME, "PERSON_LASTNAME")
css_lastname.send_keys(lastname)
css_email=driver.find_element(By.NAME, "EMAIL_ADDRESS")
css_email.send_keys(email)
# css_firstname=driver.find_element(By.NAME, "XIK_ADD_FOUND_PERSON_ID")
# add_reviewer_img = driver.find_element(By.src,"/images/en_US/buttons/create_add.gif")
add_reviewer_img = driver.find_element(By.XPATH,"//img[@src='/images/en_US/buttons/create_add.gif']")
add_reviewer_img.click()
# case 1 (exist, save and add, best case)
# add_reviewer_img = driver.find_element(By.src,"/images/en_US/buttons/create_add.gif")
success=True
try:
add_reviewer_img = driver.find_element(By.XPATH,"//img[@src='/images/en_US/buttons/save_add.gif']")
add_reviewer_img.click()
except:
success=False
# case 2 (save and send email)
# windows=driver.window_handles
# print(windows)
if not success:
try:
driver.switch_to.window(driver.window_handles[1])
iframe = driver.find_element_by_name("bottombuttons")
driver.switch_to.frame(iframe)
# add_reviewer_img = driver.find_element(By.src,"/images/en_US/buttons/create_add.gif")
add_save_send = driver.find_element(By.XPATH,"//img[@src='/images/en_US/buttons/save_send.gif']")
add_save_send.click()
success = True
except:
success = False
# case 3 # select from existing, just pick one randomly
if not success:
try:
add_save_send = driver.find_element(By.XPATH,"//img[@src='/images/en_US/buttons/add.gif']")
add_save_send.click()
success = True
except:
success = False
Below is a message from Ali Rhoades, an experiential learning coordinator.
For students interested in career readiness and job searches, I would encourage them to look into our events page including our Careerapalooza, Career Fair, and other events. Our Careerapalooza is coming up (next Thursday) but this would be a great opportunity for students to informally meet our staff, learn more about our support services (including applying for internship) and join a career community.
Our office just recently launched our Career Communities model, which encourages students to lean into finding internships and jobs specific to their passions and industry, instead of stressing to find experiential learning opportunities strict to their major. For example, if an engineering student was passionate for working in nonprofit, they would find tailored opportunities through their Nonprofit Career Community. This link will also provide them with access to their Handshake accounts (all OU students have free access to this internship and job board) including other helpful resources such as an advisor, additional job posting websites, and career readiness tips. Handshake will be the source where OU posts jobs, internships, externships, co-ops, and fellowships, which has new postings daily. If you would like one of our staff members to visit with your class to share more information about Career Communities and Handshake, please let me know-I’d be happy to set that up for you.
As for micro-internships: OU Career Center will be partnering with Parker Dewey soon, hopefully by the end of September, to have tailored micro-internships exclusively to OU students for local, remote, and regional opportunities that are short term and project based. Until then, students can look for public opportunities through Parker Dewey’s website here.
DDPG
Deep deterministic policy gradient (DDPG) is an actor-critic method. As the name suggests, the action is deterministic given the observation. DDPG composes of actor and critic networks. Given an observation, an actor network outputs an appropriate action. And given an action and an observation, a critic network outputs a prediction of the expected return (Q-value).
Replay Buffer
A replay buffer will store the (obs, action, reward, obs_next) tuples for each step the agent interacts with the environment. After the buffer is full, a batch of tuples can be extracted for training. Meanwhile, the new experience can be stored again in another buffer. Once the latter buffer is full, it can be swapped with the training buffer.
Training
A batch sampled from the replay buffer will be used to train the critic networks and then the actor networks. Then, another batch will be sampled and trained both networks again. As you will see, there are actually two critic and two actor networks in DDPG. The two networks in each type are almost identical but just one is a delayed or an average version of the other.
Training Actor Networks
We will fix the critic networks when we train the actor networks. Given a tuple (obs, action, reward, obs_next), we will only use the obs variable and plugged that into our actor network. The current actor network will create action_est and we can input both action_est and obs into the critic network to get an Q-value estimate. Assuming that the critic network is well-trained, the objective here is simply to maximize this estimated Q-value.
Training Critic Networks
Given a tuple (obs, action, reward, obs_next) and fixed actor networks, we have multiple ways to estimate an Q-value. For example, we can have
Q_est = critic_est (obs, action)
Q_tar = reward + discount 
critic_est (obs_next, act_est (obs_next))
The critic_est and act_est are critic and actor networks, respectively. So to train the network, we may try to minimize the square difference of Q_est and Q_tar.
Target Networks
As both Q_est and Q_tar depend on critic_est network, the naive implementation in training the critic networks tends to have poor convergence. To mitigate this problem, we can introduce a separate critic_est network. We will call this target critic network and denote it as critic_target. In practice, critic_target will simply be a delayed copy or an exponential average of critic_est. It seems that a target actor network, act_target, is introduced in the same manner. But I am not sure if it is really necessary.
TD3
TD3 stands for Twin Delayed DDPG. It is essentially DDPG but added with two additional tricks as follows.
Delayed Policy Update
The authors found that it is more important to have an accurate critic network than an actor network (similar to GAN that an accurate discriminator is important). Consequently, the authors suggest “delaying” the policy update. In practice, say train two batches of critic network before training a batch of actor network.
Clipped Double Q-Learning
Double Q-Learning was introduced to address the overestimation of Q-value. The authors argue that there is an overestimate of Q in DDPG as well. And they state that even double Q-learning is not sufficient to suppress the overestimation. Instead, they introduced a “clipped double Q-Learning”, where they use two critic networks rather than one to generate a current Q-estimate. And they aggregate the two estimates as the minimum of the two.