Conversational AI is revolutionizing how businesses operate in every industry with applications like virtual agents, chatbots and assistants. Creating an intelligent and intuitive app involves quickly adapting new state-of-the-art research and deploying it in production. You can learn about the latest advancements in this area at the GPU Technology Conference, taking place October 5-9.
At GTC, researchers and developers from leading institutions across the globe will share new techniques and innovations in speech recognition, natural language processing and text-to-speech technologies.
Explore these technologies in some of GTC’s conversational AI sessions, including:
- How the Revolution of Natural Language Processing Is Changing How Companies Understand Text | Hugging Face — Text is central to both individual lives and companies. Natural language processing has had its limits, from keyword-based to linguistic approaches. Learn how Hugging Face empowers companies to adopt new models faster than ever, making the libraries some of the fastest growing open-source products to date.
- Learning from Language and Interaction with Human Players in Minecraft | Facebook AI Research — CraftAssist is a research program aimed at building an assistant that can collaborate with human players in the sandbox construction game Minecraft. Learn how the platform is intended to support the study of agents and encourages the AI research community to use the platform for their own experiments.
- GPT-3: Language Models are Few-Shot Learners | OpenAI — Join a discussion GPT-3, an autoregressive language model with 175 billion parameters, which is 10x more than any previous non-sparse language model, and see how it demonstrates strong few-shot learning, sometimes even reaching competitiveness with prior state of the art fine-tuning approaches.
- Optimized Dialogue Policy Learning for a Task-Oriented Chatbot Through Deep Reinforcement Learning | PayPal — Dialogue policy manager plays a pivotal role in chatbots by guiding the conversation flow through a series of appropriate agent responses for success completion of a given task. PayPal will demonstrate how deep reinforcement learning can be used to automatically extract optimal dialogue policy from unannotated conversation logs and continuously adapt the policy to changing bot environments.
- Delivering Human AI Experience Through Voice Interactions | Voca — The availability of deep learning technologies for building voice bots, by a cascading of ASR, NLU and TTS engines, shortens the time to engineer a working production system. Join Voca as they describe their experience in building a “human like” conversational telephony-based voice bot.
- How to Build and Deploy Conversational AI Apps: A Startup Perspective | Intelligent Voice, Gong — Join two members of the NVIDIA Inception program as they share their experience with building industry-leading conversational AI solutions including using GPU-accelerated speech recognition, language understanding and text-to-speech models using NVIDIA NeMo and KALDI toolkits, as well as how to deploy them with the NVIDIA Jarvis framework.
- How to Build Enterprise-Grade AI Assistants | RASA — AI assistants are transforming business, from internal operations to banking to healthcare. RASA will discuss how to move beyond proof-of-concepts to launch production-grade AI assistants that deliver measurable value. They’ll walk through conversation-driven development, a set of practices to improve AI assistants over time and how to build something that truly helps users.
NVIDIA Deep Learning Institute at GTC
At GTC, the NVIDIA Deep Learning Institute is offering instructor-led, hands-on training on how to use Transformer-based natural language processing models for text classification tasks, such as categorizing documents.
Participants will learn how to use Transformer-based models for named-entity recognition tasks and how to analyze various model features, constraints and characteristics. The training will help developers determine which model is best suited for a particular use case based on metrics, domain specificity, and available resources.
Anyone connected online can join for a nominal fee, and participants will have access to a fully configured, GPU-accelerated server in the cloud. This in-depth, eight-hour, interactive workshop is taught by an NVIDIA subject-matter expert.