Activity
-
New research from Meta FAIR: Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model. This recipe combines next token…
New research from Meta FAIR: Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model. This recipe combines next token…
Liked by Hsiao-Ping Tseng
-
“[With Pencil Spaces] we are able to cover 50% more materials in the same amount of time. Thank you!!” Thank you, Vivian Kwek, for your wonderful…
“[With Pencil Spaces] we are able to cover 50% more materials in the same amount of time. Thank you!!” Thank you, Vivian Kwek, for your wonderful…
Liked by Hsiao-Ping Tseng
-
This week, Nvidia announced NIM Agent Blueprints, a catalog of pre-trained, customizable AI workflows for developers building and deploying…
This week, Nvidia announced NIM Agent Blueprints, a catalog of pre-trained, customizable AI workflows for developers building and deploying…
Liked by Hsiao-Ping Tseng
Experience
Education
Publications
-
Email Volume Optimization at LinkedIn
KDD 2016
In this paper, we discuss our strategy and experience with regard to the problem of email volume optimization at LinkedIn. In particular, we present a cost-benefit analysis of sending emails, the key factors to administer an effective volume optimization, our algorithm for volume optimization, the architecture of the supporting system, and experimental results from online A/B tests.
Other authors -
Personalizing LinkedIn Feed
KDD 2015
LinkedIn dynamically delivers update activities from a user's interpersonal network to more than 300 million members in the personalized feed that ranks activities according their "relevance" to the user. This paper discloses the implementation details behind this personalized feed system at LinkedIn which can not be found from related work, and addresses the scalability and data sparsity challenges for deploying the system online. More specifically, we focus on the personalization models by…
LinkedIn dynamically delivers update activities from a user's interpersonal network to more than 300 million members in the personalized feed that ranks activities according their "relevance" to the user. This paper discloses the implementation details behind this personalized feed system at LinkedIn which can not be found from related work, and addresses the scalability and data sparsity challenges for deploying the system online. More specifically, we focus on the personalization models by generating three kinds of affinity scores: Viewer-ActivityType Affinity, Viewer-Actor Affinity, and Viewer-Actor-ActivityType Affinity. Extensive experiments based on online bucket tests (A/B experiments) and offline evaluation illustrate the effect of our personalization models in LinkedIn feed.
Other authors
Patents
-
Content Generation And Targeting Using Machine Learning
Issued US 10,671,680
-
Techniques for Expanding a Target Audience for Messaging
Issued US 10643226
-
MANAGING MULTIPLE MESSAGES
Issued US 10,263,941
-
Creation and delivery of individually customized web pages
Issued US 10122824
A system includes a computer storage device to store a first and second set of information about individuals, and user interface design components. A computer server is coupled to the computer storage and is programmed to receive a request for a webpage from an individual's device and analyze the first and second set of information to predict differences in at least one key performance indicator for that individual. The server is further programmed to automatically select a user interface…
A system includes a computer storage device to store a first and second set of information about individuals, and user interface design components. A computer server is coupled to the computer storage and is programmed to receive a request for a webpage from an individual's device and analyze the first and second set of information to predict differences in at least one key performance indicator for that individual. The server is further programmed to automatically select a user interface design component for presentation to the individual, automatically generate and transmit a first customized webpage with the selected user interface design component to the device, and in response to an interaction with the selected user interface design component on the first customized webpage, automatically generate and transmit a second customized webpage to the device to create a customized page flow.
-
Second-Pass Ranker for Push Notifications in a Social Network
Issued US 9,413,838
Machine learning based methods and systems for selecting optimal content for push notification from a heterogenous pool of content.
Other inventorsSee patent -
Image manipulation for web content
Issued US 9,405,734
A web page or group of pages can be automatically generated based on images selected by a user. The images are analyzed for color and groups of colors for the page background and image framing are determined. Images are analyzed to determine the shapes of the images and the images are placed within appropriate shapes on a page on the web.
Other inventorsSee patent -
Semiconductor hold time fixing
Issued US 9,171,112
Computer implemented techniques are disclosed for fixing signal hold-time violations in semiconductor chips. Analysis includes estimation of hold-time requirements using ideal clocks. Allocation of placement regions within the design and near the macro circuits allows for later placement and wiring use during layout hold-time fixing.
Other inventorsSee patent -
Multi-threaded global routing
Issued US US 8,464,197
-
Proactive Identification of content items for a member of a social network
Filed US 3080.D35PRV
A system and method for behavior influenced search ranking may include obtaining, via a network interface, a search term from a user device. An initial result including a first group of the user profiles may be generated based on user profiles from a social network in relation to the search term, the user profiles stored in a profile database. A rank of each of the first group of the user profiles may be determined based, at least in part, on interactions from an activity database corresponding…
A system and method for behavior influenced search ranking may include obtaining, via a network interface, a search term from a user device. An initial result including a first group of the user profiles may be generated based on user profiles from a social network in relation to the search term, the user profiles stored in a profile database. A rank of each of the first group of the user profiles may be determined based, at least in part, on interactions from an activity database corresponding to the first group of the user profiles, the activity database storing information indicative of activities related to the social network, the activities including the interactions. The user device may display a second group of the user profiles, including at least some of the first group of the user profiles, according to the rank of the first group of the user profiles.
Other inventors
Projects
More activity by Hsiao-Ping
-
Lately, there’s been a growing chorus claiming Generative AI is the next overhyped tech darling. Critics are lining up to call it just another bubble…
Lately, there’s been a growing chorus claiming Generative AI is the next overhyped tech darling. Critics are lining up to call it just another bubble…
Liked by Hsiao-Ping Tseng
-
My team at Otter.ai just launched some new features to help make your meetings more efficient. Earlier this year, we added Action Items, that uses AI…
My team at Otter.ai just launched some new features to help make your meetings more efficient. Earlier this year, we added Action Items, that uses AI…
Liked by Hsiao-Ping Tseng
-
Launching one product at a time feels too slow in the AI era, our community is asking for more. Today we’re excited to introduce two new ways to…
Launching one product at a time feels too slow in the AI era, our community is asking for more. Today we’re excited to introduce two new ways to…
Liked by Hsiao-Ping Tseng
-
🥾🍽️ Hiking + Lunch = Perfect Team Day! 🍽️🥾 Last week, we traded our desks for the great outdoors and went on a beautiful hike through the Alpine…
🥾🍽️ Hiking + Lunch = Perfect Team Day! 🍽️🥾 Last week, we traded our desks for the great outdoors and went on a beautiful hike through the Alpine…
Liked by Hsiao-Ping Tseng
-
If you are going to swampUP, let me know!
If you are going to swampUP, let me know!
Liked by Hsiao-Ping Tseng
-
Our team at Facebook Feed is currently looking to fill an IC5 Machine Learning Engineer position, in Menlo Park, California. If you are passionate…
Our team at Facebook Feed is currently looking to fill an IC5 Machine Learning Engineer position, in Menlo Park, California. If you are passionate…
Liked by Hsiao-Ping Tseng
-
🚀 What a great week at #KDD in Barcelona! We were thrilled to see Hema Raghavan and Matthias Fey take the stage, diving deep into the future of…
🚀 What a great week at #KDD in Barcelona! We were thrilled to see Hema Raghavan and Matthias Fey take the stage, diving deep into the future of…
Liked by Hsiao-Ping Tseng
-
I felt truly energized to see so many young and eager-to-learn faces at the first lecture of my cross-listed undergraduate/graduate course on…
I felt truly energized to see so many young and eager-to-learn faces at the first lecture of my cross-listed undergraduate/graduate course on…
Liked by Hsiao-Ping Tseng
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More