IBM Booth @ NeurIPS Schedule
Report abuse
Use this data
Sign up for free
Schedule Slot Title
Day
12/11 Mon
Count11
1
BOOTH | Demo: MoLFormers: Foundation Models for Chemistry
2
BOOTH | Demo: A Platform For Auditing the Trustworthiness of Synthetic Data
3
BOOTH | Demo: Industrial demonstration of popular backbones for Time series Foundation Models
4
BOOTH | Demo: FlowPilot: An LLM-Powered System for Enterprise Data Integration
5
BOOTH | Demo: The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
6
BOOTH | Demo: MoLFormers: Foundation Models for Chemistry
7
BOOTH | Demo: Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
8
BOOTH | Demo: RADAR: A Robust AI-Written Text Detector
9
BOOTH | Demo: AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
10
BOOTH | <no demo>
11
BOOTH | <no demo>
12
BOOTH | Demo: AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
13
BOOTH | Demo: A Platform For Auditing the Trustworthiness of Synthetic Data
14
BOOTH | Demo: MoLFormers: Foundation Models for Chemistry
15
BOOTH | Demo: The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
16
BOOTH | Demo: FlowPilot: An LLM-Powered System for Enterprise Data Integration
17
BOOTH | Demo: Industrial demonstration of popular backbones for Time series Foundation Models
18
BOOTH | Demo: RADAR: A Robust AI-Written Text Detector
19
BOOTH | Demo: Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
20
BOOTH | Demo: AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
21
BOOTH | Demo: FlowPilot: An LLM-Powered System for Enterprise Data Integration
22
BOOTH | Demo: MoLFormers: Foundation Models for Chemistry
23
BOOTH | Demo: Industrial demonstration of popular backbones for Time series Foundation Models
24
BOOTH | Demo: RADAR: A Robust AI-Written Text Detector
25
BOOTH | Demo: A Platform For Auditing the Trustworthiness of Synthetic Data
26
BOOTH | Demo: The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
27
BOOTH | Demo: Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
28
BOOTH | Demo: Industrial demonstration of popular backbones for Time series Foundation Models
29
BOOTH | Demo: The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
30
BOOTH | Demo: Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
31
BOOTH | Demo: FlowPilot: An LLM-Powered System for Enterprise Data Integration
32
BOOTH | Demo: RADAR: A Robust AI-Written Text Detector
Day
12/12 Tue
Count8
Day
12/13 Wed
Count8
Day
12/14 Thu
Count5
Drag to adjust the number of frozen columns
Time
Demo Title
Demo Presenter(s)
Demo Abstract
Booth Duty: Research Staff
9:00 AM - 10:00 AM
MoLFormers: Foundation Models for Chemistry
Samuel Hoffman
Hendrik Strobelt

MoLFormers is a suite of molecular foundation models that accelerate the development of chemicals, materials, and biologics. It includes a number of state-of-the-art pre-trained and fine-tuned models for molecular data analysis, property prediction, nearest neighbor retrieval, and discovery (such as design of catalysts, stability agents, antibodies, protein inhibitors) tasks, thus enabling molecular science at higher accuracy and success rate, lower cost, and faster speed. The suite contains high-quality families (encoder-only, encoder-decoder, decoder-only) of large molecular models that combine best-of-breed architectures (e.g., rotary embedding, multi-modal embedding, prompt tuning, linear attention, geometry encoder, etc.) and provide world-class predictive and generative AI capabilities. One such instantiation of MoLFormer is MoLFormer-XL, a chemical SMILES encoder trained on 1.1 B chemicals using rotary positional encodings and linear attention. MoLFormer-XL, finetuned on downstr

Hao Wang
Vojtech Havlicek
Jannis Born
Zhenfang Chen
Abhishek Bhandwaldar
10:00 AM - 11:00 AM
A Platform For Auditing the Trustworthiness of Synthetic Data
Youssef Mroueh

With the accelerated advances in Generative AI, Gartner predicted that by 2030, synthetic data will overshadow real data in training AI models. The primary motivation for using synthetic data in the AI life cycle comes from its promise in alleviating risk, bias, harm, and privacy concerns inherent in the real data. How can we guarantee that such an approach delivers on its promises? How can we determine that a synthetic dataset indeed conforms to fairness, privacy, robustness and fidelity requirements, and is still useful to downstream tasks?


Radu Marinescu
Michael Katz
Amit Dhurandhar
Vojtech Havlicek
Sivan Doveh
11:00 AM - 12:00 PM
Industrial demonstration of popular backbones for Time series Foundation Models
Nam Nguyen

While Foundation Models (FM) have greatly transformed AI solutions for language and vision, they often fall short in addressing time-series data, which is widely used in various industries. At IBM Research, our dedicated team focuses exclusively on advancing Time Series Foundation Models and has made significant contributions with several influential papers presented at top AI conferences. Our team has been pioneers in this space where we defined the first inaugural architecture for several popular Time-series FM backbones, including the first transformer for multi-variate time-series representation learning (TST, KDD 21), the first patched time-series transformer (PatchTST, ICLR 23) and the first patched MLP-Mixer for time series (PatchTSMixer, KDD 23). Our latest Models (PatchTST and PatchTSMixer) are the leading SOTAs in this space with a significant reduction (2-3X) in compute and memory requirements. We have released our SOTA models through various open-source channels attracting

Benjamin Hoover
Michael Katz
Igor Melnyk
Kavitha Srinivas
Benedikt Blumenstiel
12:00 PM - 1:00 PM
FlowPilot: An LLM-Powered System for Enterprise Data Integration
Enrico Toniato

Traditional data integration techniques often require complex coding and a deep understanding of data architectures, which can be daunting for non-specialists. In the evolving landscape of AI, there's a growing need for tools that democratize data access and analysis. We present FlowPilot, a novel system that departs from the current one-shot text-to-SQL paradigms that often fail to answer complex queries.


Vojtech Havlicek
Nikita Janakarajan
Erik Miehling
Svetoslav Nizhnichenkov
Kavitha Srinivas
1:00 PM - 2:00 PM
The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
Maya Murad

The emergence of foundation models has significantly lowered the barriers to applying AI to everyday problems, transforming the way organizations consume, customize and build AI-enabled applications. We are also seeing the emergence of a new persona, the AI Builder, who is in need of dedicated tooling to harness the power of LLMs while mitigating its associated risks.


Hao Wang
Payel Das
Nikita Janakarajan
Zhenfang Chen
Kavitha Srinivas
Rahul Krishna
2:00 PM - 3:00 PM
MoLFormers: Foundation Models for Chemistry
Samuel Hoffman
Hendrik Strobelt

MoLFormers is a suite of molecular foundation models that accelerate the development of chemicals, materials, and biologics. It includes a number of state-of-the-art pre-trained and fine-tuned models for molecular data analysis, property prediction, nearest neighbor retrieval, and discovery (such as design of catalysts, stability agents, antibodies, protein inhibitors) tasks, thus enabling molecular science at higher accuracy and success rate, lower cost, and faster speed. The suite contains high-quality families (encoder-only, encoder-decoder, decoder-only) of large molecular models that combine best-of-breed architectures (e.g., rotary embedding, multi-modal embedding, prompt tuning, linear attention, geometry encoder, etc.) and provide world-class predictive and generative AI capabilities. One such instantiation of MoLFormer is MoLFormer-XL, a chemical SMILES encoder trained on 1.1 B chemicals using rotary positional encodings and linear attention. MoLFormer-XL, finetuned on downstr

Hao Wang
Lam Nguyen
Sivan Doveh
Leonid Karlinsky
Kavitha Srinivas
3:00 PM - 4:00 PM
Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
Yu Deng
Saurabh Jha

The fast-increasing complexity of modern IT in multi cloud environments is bringing unprecedented management challenges to Site Reliability Engineers (SREs) to meet Service Level Objectives (SLOs) and keep systems up and running effectively. To put in perspective, an availability SLO of 99.99% allows for 4.3 minutes of downtime per month, hardly something that can be attained by simply reacting to incidents. In this demo, we introduce our approach to address this challenge by transforming ITOps from being reactive to becoming proactive by leveraging large language models and advanced AI capabilities. The main goal of our work is to automate as much as possible the implementation of resolutions for upcoming IT issues before they turn into outages. Our demo consists of four steps:

(1) Issue Detection, where we have developed an unsupervised methodology for detecting issues via ensemble of various anomaly detectors. We compare our methods with the state-of-the-art techniques implemented i

Benjamin Hoover
Lam Nguyen
Jannis Born
Leonid Karlinsky
Trey Tinnell
4:00 PM - 5:00 PM
RADAR: A Robust AI-Written Text Detector
Pin-Yu Chen
Hendrik Strobelt

This demo shows RADAR, a language model trained to detect AI-written text generated from large language models (LLMs). The live demo consists of four RADAR detectors trained on different LLMs. Users can paste in a paragraph of text and receive likelihood predictions of the AI-written text in real time. Examples of AI versus human written text are also provided in the demo. The demo is based on the NeurIPS 2023 paper "RADAR: Robust AI-Text Detection via Adversarial Learning".  


Benjamin Hoover
Miao Liu
Benedikt Blumenstiel
Igor Melnyk
Junkyu Lee
5:00 PM - 6:00 PM
AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
Nathalie Baracaldo

With the wide range of new applications using machine learning models, and the recent disruptive power of foundation models, the robustness and trustworthiness of artificial intelligence has become imperative. We will present the Adversarial Robustness Toolbox (ART), an open-source project which is part of The Linux Foundation AI & Data, that provides tools to assess and improve the security of machine learning models. Recently, ART supports for models hosted by Hugging Face, and as such, we will present a demonstration of how to use this new functionality.

Dennis Wei
Emilio Vital Brazil
Debarun Bhattacharjya
Sivan Doveh
Daniel Salles Civitarese
6:00 PM - 7:00 PM
Dennis Wei
Debarun Bhattacharjya
Apoorva Nitsure
Svetoslav Nizhnichenkov
Oliver Schilter
7:00 PM - 8:30 PM
Karthikeyan Natesan Ramamurthy
Svetoslav Nizhnichenkov
Nima Dehmamy
Daniel Salles Civitarese
Rahul Krishna
9:00 AM - 10:00 AM
AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
Nathalie Baracaldo

With the wide range of new applications using machine learning models, and the recent disruptive power of foundation models, the robustness and trustworthiness of artificial intelligence has become imperative. We will present the Adversarial Robustness Toolbox (ART), an open-source project which is part of The Linux Foundation AI & Data, that provides tools to assess and improve the security of machine learning models. Recently, ART supports for models hosted by Hugging Face, and as such, we will present a demonstration of how to use this new functionality.

Miao Liu
Kristjan Greenewald
Leshem Choshen
Braulio Dumba
Ruchi Mahindru
10:00 AM - 11:00 AM
A Platform For Auditing the Trustworthiness of Synthetic Data
Youssef Mroueh

With the accelerated advances in Generative AI, Gartner predicted that by 2030, synthetic data will overshadow real data in training AI models. The primary motivation for using synthetic data in the AI life cycle comes from its promise in alleviating risk, bias, harm, and privacy concerns inherent in the real data. How can we guarantee that such an approach delivers on its promises? How can we determine that a synthetic dataset indeed conforms to fairness, privacy, robustness and fidelity requirements, and is still useful to downstream tasks?


Radu Marinescu
Kristjan Greenewald
Leshem Choshen
Eduardo Soares
Erik Miehling
Abhishek Bhandwaldar
11:00 AM - 12:00 PM
MoLFormers: Foundation Models for Chemistry
Samuel Hoffman
Hendrik Strobelt

MoLFormers is a suite of molecular foundation models that accelerate the development of chemicals, materials, and biologics. It includes a number of state-of-the-art pre-trained and fine-tuned models for molecular data analysis, property prediction, nearest neighbor retrieval, and discovery (such as design of catalysts, stability agents, antibodies, protein inhibitors) tasks, thus enabling molecular science at higher accuracy and success rate, lower cost, and faster speed. The suite contains high-quality families (encoder-only, encoder-decoder, decoder-only) of large molecular models that combine best-of-breed architectures (e.g., rotary embedding, multi-modal embedding, prompt tuning, linear attention, geometry encoder, etc.) and provide world-class predictive and generative AI capabilities. One such instantiation of MoLFormer is MoLFormer-XL, a chemical SMILES encoder trained on 1.1 B chemicals using rotary positional encodings and linear attention. MoLFormer-XL, finetuned on downstr

Amit Dhurandhar
Vojtech Havlicek
Igor Melnyk
Ruchi Mahindru
Junkyu Lee
12:00 PM - 1:00 PM
The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
Maya Murad

The emergence of foundation models has significantly lowered the barriers to applying AI to everyday problems, transforming the way organizations consume, customize and build AI-enabled applications. We are also seeing the emergence of a new persona, the AI Builder, who is in need of dedicated tooling to harness the power of LLMs while mitigating its associated risks.


Jannis Born
Nikita Janakarajan
Hiroshi Kajino
Karthikeyan Natesan Ramamurthy
Svetoslav Nizhnichenkov
Trey Tinnell
1:00 PM - 2:00 PM
FlowPilot: An LLM-Powered System for Enterprise Data Integration
Enrico Toniato

Traditional data integration techniques often require complex coding and a deep understanding of data architectures, which can be daunting for non-specialists. In the evolving landscape of AI, there's a growing need for tools that democratize data access and analysis. We present FlowPilot, a novel system that departs from the current one-shot text-to-SQL paradigms that often fail to answer complex queries.


Songtao Lu
Hiroshi Kajino
Oliver Schilter
Karthikeyan Natesan Ramamurthy
Zhenfang Chen
Trey Tinnell
2:00 PM - 3:00 PM
Industrial demonstration of popular backbones for Time series Foundation Models
Nam Nguyen

While Foundation Models (FM) have greatly transformed AI solutions for language and vision, they often fall short in addressing time-series data, which is widely used in various industries. At IBM Research, our dedicated team focuses exclusively on advancing Time Series Foundation Models and has made significant contributions with several influential papers presented at top AI conferences. Our team has been pioneers in this space where we defined the first inaugural architecture for several popular Time-series FM backbones, including the first transformer for multi-variate time-series representation learning (TST, KDD 21), the first patched time-series transformer (PatchTST, ICLR 23) and the first patched MLP-Mixer for time series (PatchTSMixer, KDD 23). Our latest Models (PatchTST and PatchTSMixer) are the leading SOTAs in this space with a significant reduction (2-3X) in compute and memory requirements. We have released our SOTA models through various open-source channels attracting

Benedikt Blumenstiel
Debarun Bhattacharjya
Songtao Lu
Shivchander Sudalairaj
Oliver Schilter
Zhenfang Chen
3:00 PM - 4:00 PM
RADAR: A Robust AI-Written Text Detector
Pin-Yu Chen
Hendrik Strobelt

This demo shows RADAR, a language model trained to detect AI-written text generated from large language models (LLMs). The live demo consists of four RADAR detectors trained on different LLMs. Users can paste in a paragraph of text and receive likelihood predictions of the AI-written text in real time. Examples of AI versus human written text are also provided in the demo. The demo is based on the NeurIPS 2023 paper "RADAR: Robust AI-Text Detection via Adversarial Learning".  


Dmitry Krotov
Dennis Wei
Hiroshi Kajino
Shivchander Sudalairaj
Apoorva Nitsure
Michael Katz
4:00 PM - 5:00 PM
Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
Yu Deng
Saurabh Jha

The fast-increasing complexity of modern IT in multi cloud environments is bringing unprecedented management challenges to Site Reliability Engineers (SREs) to meet Service Level Objectives (SLOs) and keep systems up and running effectively. To put in perspective, an availability SLO of 99.99% allows for 4.3 minutes of downtime per month, hardly something that can be attained by simply reacting to incidents. In this demo, we introduce our approach to address this challenge by transforming ITOps from being reactive to becoming proactive by leveraging large language models and advanced AI capabilities. The main goal of our work is to automate as much as possible the implementation of resolutions for upcoming IT issues before they turn into outages. Our demo consists of four steps:

(1) Issue Detection, where we have developed an unsupervised methodology for detecting issues via ensemble of various anomaly detectors. We compare our methods with the state-of-the-art techniques implemented i

Parikshit Ram
Miao Liu
Payel Das
Dmitry Krotov
Tim Rumbell
9:00 AM - 10:00 AM
AI Robustness in the Era of Foundation Models: State of the ART and Open Challenges
Nathalie Baracaldo

With the wide range of new applications using machine learning models, and the recent disruptive power of foundation models, the robustness and trustworthiness of artificial intelligence has become imperative. We will present the Adversarial Robustness Toolbox (ART), an open-source project which is part of The Linux Foundation AI & Data, that provides tools to assess and improve the security of machine learning models. Recently, ART supports for models hosted by Hugging Face, and as such, we will present a demonstration of how to use this new functionality.

Leshem Choshen
Eduardo Soares
Emilio Vital Brazil
Djalel Bouneffouf
Nima Dehmamy
10:00 AM - 11:00 AM
FlowPilot: An LLM-Powered System for Enterprise Data Integration
Enrico Toniato

Traditional data integration techniques often require complex coding and a deep understanding of data architectures, which can be daunting for non-specialists. In the evolving landscape of AI, there's a growing need for tools that democratize data access and analysis. We present FlowPilot, a novel system that departs from the current one-shot text-to-SQL paradigms that often fail to answer complex queries.


Kristjan Greenewald
Amit Dhurandhar
Erik Miehling
Ruchi Mahindru
Leonid Karlinsky
11:00 AM - 12:00 PM
MoLFormers: Foundation Models for Chemistry
Samuel Hoffman
Hendrik Strobelt

MoLFormers is a suite of molecular foundation models that accelerate the development of chemicals, materials, and biologics. It includes a number of state-of-the-art pre-trained and fine-tuned models for molecular data analysis, property prediction, nearest neighbor retrieval, and discovery (such as design of catalysts, stability agents, antibodies, protein inhibitors) tasks, thus enabling molecular science at higher accuracy and success rate, lower cost, and faster speed. The suite contains high-quality families (encoder-only, encoder-decoder, decoder-only) of large molecular models that combine best-of-breed architectures (e.g., rotary embedding, multi-modal embedding, prompt tuning, linear attention, geometry encoder, etc.) and provide world-class predictive and generative AI capabilities. One such instantiation of MoLFormer is MoLFormer-XL, a chemical SMILES encoder trained on 1.1 B chemicals using rotary positional encodings and linear attention. MoLFormer-XL, finetuned on downstr

Parikshit Ram
Radu Marinescu
Michael Katz
Benedikt Blumenstiel
Igor Melnyk
Junkyu Lee
12:00 PM - 1:00 PM
Industrial demonstration of popular backbones for Time series Foundation Models
Nam Nguyen

While Foundation Models (FM) have greatly transformed AI solutions for language and vision, they often fall short in addressing time-series data, which is widely used in various industries. At IBM Research, our dedicated team focuses exclusively on advancing Time Series Foundation Models and has made significant contributions with several influential papers presented at top AI conferences. Our team has been pioneers in this space where we defined the first inaugural architecture for several popular Time-series FM backbones, including the first transformer for multi-variate time-series representation learning (TST, KDD 21), the first patched time-series transformer (PatchTST, ICLR 23) and the first patched MLP-Mixer for time series (PatchTSMixer, KDD 23). Our latest Models (PatchTST and PatchTSMixer) are the leading SOTAs in this space with a significant reduction (2-3X) in compute and memory requirements. We have released our SOTA models through various open-source channels attracting

Payel Das
Jannis Born
Abbas Rahimi
Mikhail Yurochkin
Ruchi Mahindru
1:00 PM - 2:00 PM
RADAR: A Robust AI-Written Text Detector
Pin-Yu Chen
Hendrik Strobelt

This demo shows RADAR, a language model trained to detect AI-written text generated from large language models (LLMs). The live demo consists of four RADAR detectors trained on different LLMs. Users can paste in a paragraph of text and receive likelihood predictions of the AI-written text in real time. Examples of AI versus human written text are also provided in the demo. The demo is based on the NeurIPS 2023 paper "RADAR: Robust AI-Text Detection via Adversarial Learning".  


Abbas Rahimi
Flaviu Cipcigan
Apoorva Nitsure
Nima Dehmamy
Djalel Bouneffouf
Trey Tinnell
2:00 PM - 3:00 PM
A Platform For Auditing the Trustworthiness of Synthetic Data
Youssef Mroueh

With the accelerated advances in Generative AI, Gartner predicted that by 2030, synthetic data will overshadow real data in training AI models. The primary motivation for using synthetic data in the AI life cycle comes from its promise in alleviating risk, bias, harm, and privacy concerns inherent in the real data. How can we guarantee that such an approach delivers on its promises? How can we determine that a synthetic dataset indeed conforms to fairness, privacy, robustness and fidelity requirements, and is still useful to downstream tasks?


Flaviu Cipcigan
Songtao Lu
Shivchander Sudalairaj
Karthikeyan Natesan Ramamurthy
Apoorva Nitsure
Nima Dehmamy
3:00 PM - 4:00 PM
The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
Maya Murad

The emergence of foundation models has significantly lowered the barriers to applying AI to everyday problems, transforming the way organizations consume, customize and build AI-enabled applications. We are also seeing the emergence of a new persona, the AI Builder, who is in need of dedicated tooling to harness the power of LLMs while mitigating its associated risks.


Erik Miehling
Emilio Vital Brazil
Debarun Bhattacharjya
Hiroshi Kajino
Abhishek Bhandwaldar
4:00 PM - 5:00 PM
Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
Yu Deng
Saurabh Jha

The fast-increasing complexity of modern IT in multi cloud environments is bringing unprecedented management challenges to Site Reliability Engineers (SREs) to meet Service Level Objectives (SLOs) and keep systems up and running effectively. To put in perspective, an availability SLO of 99.99% allows for 4.3 minutes of downtime per month, hardly something that can be attained by simply reacting to incidents. In this demo, we introduce our approach to address this challenge by transforming ITOps from being reactive to becoming proactive by leveraging large language models and advanced AI capabilities. The main goal of our work is to automate as much as possible the implementation of resolutions for upcoming IT issues before they turn into outages. Our demo consists of four steps:

(1) Issue Detection, where we have developed an unsupervised methodology for detecting issues via ensemble of various anomaly detectors. We compare our methods with the state-of-the-art techniques implemented i

Parikshit Ram
Payel Das
Abbas Rahimi
Tim Rumbell
9:00 AM - 10:00 AM
Industrial demonstration of popular backbones for Time series Foundation Models
Nam Nguyen

While Foundation Models (FM) have greatly transformed AI solutions for language and vision, they often fall short in addressing time-series data, which is widely used in various industries. At IBM Research, our dedicated team focuses exclusively on advancing Time Series Foundation Models and has made significant contributions with several influential papers presented at top AI conferences. Our team has been pioneers in this space where we defined the first inaugural architecture for several popular Time-series FM backbones, including the first transformer for multi-variate time-series representation learning (TST, KDD 21), the first patched time-series transformer (PatchTST, ICLR 23) and the first patched MLP-Mixer for time series (PatchTSMixer, KDD 23). Our latest Models (PatchTST and PatchTSMixer) are the leading SOTAs in this space with a significant reduction (2-3X) in compute and memory requirements. We have released our SOTA models through various open-source channels attracting

Kristjan Greenewald
Leshem Choshen
Eduardo Soares
Dennis Wei
Emilio Vital Brazil
10:00 AM - 11:00 AM
The BAM Laboratory - Empowering AI Builders with User-Centric Tooling
Maya Murad

The emergence of foundation models has significantly lowered the barriers to applying AI to everyday problems, transforming the way organizations consume, customize and build AI-enabled applications. We are also seeing the emergence of a new persona, the AI Builder, who is in need of dedicated tooling to harness the power of LLMs while mitigating its associated risks.


Miao Liu
Radu Marinescu
Amit Dhurandhar
Eduardo Soares
Mikhail Yurochkin
11:00 AM - 12:00 PM
Detection Diagnosis and Remediation for IT Incidents powered by Generative AI
Yu Deng
Saurabh Jha

The fast-increasing complexity of modern IT in multi cloud environments is bringing unprecedented management challenges to Site Reliability Engineers (SREs) to meet Service Level Objectives (SLOs) and keep systems up and running effectively. To put in perspective, an availability SLO of 99.99% allows for 4.3 minutes of downtime per month, hardly something that can be attained by simply reacting to incidents. In this demo, we introduce our approach to address this challenge by transforming ITOps from being reactive to becoming proactive by leveraging large language models and advanced AI capabilities. The main goal of our work is to automate as much as possible the implementation of resolutions for upcoming IT issues before they turn into outages. Our demo consists of four steps:

(1) Issue Detection, where we have developed an unsupervised methodology for detecting issues via ensemble of various anomaly detectors. We compare our methods with the state-of-the-art techniques implemented i

Michael Katz
Dmitry Krotov
Mikhail Yurochkin
Junkyu Lee
Djalel Bouneffouf
12:00 PM - 1:00 PM
FlowPilot: An LLM-Powered System for Enterprise Data Integration
Enrico Toniato

Traditional data integration techniques often require complex coding and a deep understanding of data architectures, which can be daunting for non-specialists. In the evolving landscape of AI, there's a growing need for tools that democratize data access and analysis. We present FlowPilot, a novel system that departs from the current one-shot text-to-SQL paradigms that often fail to answer complex queries.


Dmitry Krotov
Nikita Janakarajan
Abbas Rahimi
Flaviu Cipcigan
Mikhail Yurochkin
1:00 PM - 2:30 PM
RADAR: A Robust AI-Written Text Detector
Pin-Yu Chen
Hendrik Strobelt

This demo shows RADAR, a language model trained to detect AI-written text generated from large language models (LLMs). The live demo consists of four RADAR detectors trained on different LLMs. Users can paste in a paragraph of text and receive likelihood predictions of the AI-written text in real time. Examples of AI versus human written text are also provided in the demo. The demo is based on the NeurIPS 2023 paper "RADAR: Robust AI-Text Detection via Adversarial Learning".  


Parikshit Ram
Dan Cunnington
Flaviu Cipcigan
Daniel Salles Civitarese
32 records

Alert

Lorem ipsum
Okay