This document discusses streaming analytics and how traditional machine learning algorithms are not well-suited for streaming data. It introduces Hierarchical Temporal Memory (HTM) as a new approach inspired by neuroscience that can handle streaming data, continuous learning, and temporal modeling. HTM uses sparse distributed representations and models sequences to make predictions and detect anomalies. The document provides examples of how HTM can be applied to problems like anomaly detection in server metrics, human behavior, geospatial tracking, social media streams, and stock prices. HTM algorithms are domain-independent and use the same codebase and parameters across different problem types.
Abstract:
There’s no question that we are seeing an increase in the availability of streaming, time-series data. Largely driven by the rise of the Internet of Things (IoT) and connected real-time data sources, we now have an enormous number of applications with sensors that produce important data that changes over time. This data presents a challenge and opportunity for businesses across every industry. How do they handle the onslaught of streaming data? How can they exploit it to make decisions in real-time? One way is to detect, in real time, when something unusual occurs. Early anomaly detection in streaming data has significant implications, yet can be very difficult to execute. It requires detectors to process data in real-time, not batches, and learn while simultaneously making predictions. In this talk, we’ll look at algorithms designed for such data and analyze the components that lead to optimal performance. We’ll also discuss a new benchmark with a labeled, real-world data set, designed to provide a controlled and repeatable environment of open-source tools to test and measure anomaly detection algorithms on streaming data. How do we score in a way that rewards algorithms that detect all anomalies as soon as possible, triggers no false alarms, works with real-world time-series data across a variety of domains, and automatically adapts to changing statistics?
Evaluating Real-Time Anomaly Detection: The Numenta Anomaly BenchmarkNumenta
Subutai Ahmad, VP Research presenting NAB and discussing the need for evaluating real-time anomaly detection algorithms. This presentation was delivered at MLConf (Machine Learning Conference) in San Francisco 2015.
Extending Flink for anomaly detection with Hierarchical Temporal Memory (HTM). Presented at Bay Area Apache Flink Meetup, in San Jose on June 27, 2016.
https://github.com/htm-community/flink-htm
Abstract:
There’s no question that we are seeing an increase in the availability of streaming, time-series data. Largely driven by the rise of the Internet of Things (IoT) and connected real-time data sources, we now have an enormous number of applications with sensors that produce important data that changes over time. This data presents a challenge and opportunity for businesses across every industry. How do they handle the onslaught of streaming data? How can they exploit it to make decisions in real-time? One way is to detect, in real time, when something unusual occurs. Early anomaly detection in streaming data has significant implications, yet can be very difficult to execute. It requires detectors to process data in real-time, not batches, and learn while simultaneously making predictions. In this talk, we’ll look at algorithms designed for such data and analyze the components that lead to optimal performance. We’ll also discuss a new benchmark with a labeled, real-world data set, designed to provide a controlled and repeatable environment of open-source tools to test and measure anomaly detection algorithms on streaming data. How do we score in a way that rewards algorithms that detect all anomalies as soon as possible, triggers no false alarms, works with real-world time-series data across a variety of domains, and automatically adapts to changing statistics?
Evaluating Real-Time Anomaly Detection: The Numenta Anomaly BenchmarkNumenta
Subutai Ahmad, VP Research presenting NAB and discussing the need for evaluating real-time anomaly detection algorithms. This presentation was delivered at MLConf (Machine Learning Conference) in San Francisco 2015.
Extending Flink for anomaly detection with Hierarchical Temporal Memory (HTM). Presented at Bay Area Apache Flink Meetup, in San Jose on June 27, 2016.
https://github.com/htm-community/flink-htm
Exploration of U-Net and Support Vector Machine classification methods for UAV multispectral image segmentation
Recently, many solutions have been introduced to accurately and automatically analyze data acquired with Unmanned Aerial Vehicles (UAVs), in particular by relying on algorithms based on Artificial Intelligence (AI) techniques. Among these, the most popular are those belonging to the category of neural networks. These techniques allow the development of ad-hoc and end-to-end solutions for the classification and segmentation of different object categories through the analysis of high-resolution multispectral images. In our research, two main methodologies have been explored for the automatic segmentation of crop rows from multispectral images acquired with UAVs. The first is based on Support Vector Machines, know to handle well overfitting issues, and the other through the implementation of “U-Net”, a state-of-the-art Convolution Neural Network
Anomaly detection in real-time data streams using HeronArun Kejariwal
Twitter has become the de facto medium for consumption of news in real time, and billions of events are generated and analyzed on a daily basis. To analyze these events, Twitter designed its own next-generation streaming system, Heron. Arun Kejariwal and Karthik Ramasamy walk you through how Heron is used to detect anomalies in real-time data streams. Although there’s been over 75 years of prior work in anomaly detection, most of the techniques cannot be used off the shelf because they’re not suitable for high-velocity data streams. Arun and Karthik explain how to make trade-offs between accuracy and speed and discuss incremental approaches that marry sampling with robust measures such as median and MCD for anomaly detection.
Energy Monitoring With Self-taught Deep NetworkYiqun Hu
This is the presentation of my talk in O'Reilly Strata Data Conference Singapore 2017. It is about how we can extract useful knowledge from unlabelled time series to help energy monitoring applications.
Data Data Everywhere: Not An Insight to Take Action UponArun Kejariwal
The big data era is characterized by ever-increasing velocity and volume of data. Over the last two or three years, several talks at Velocity have explored how to analyze operations data at scale, focusing on anomaly detection, performance analysis, and capacity planning, to name a few topics. Knowledge sharing of the techniques for the aforementioned problems helps the community to build highly available, performant, and resilient systems.
A key aspect of operations data is that data may be missing—referred to as “holes”—in the time series. This may happen for a wide variety of reasons, including (but not limited to):
# Packets being dropped due to unresponsive downstream services
# A network hiccup
# Transient hardware or software failure
# An issue with the data collection service
“Holes” in the time series on data analysis can potentially skew the analysis of data. This in turn can materially impact decision making. Arun Kejariwal presents approaches for analyzing operations data in the presence of “holes” in the time series, highlighting how missing data impacts common data analysis such as anomaly detection and forecasting, discussing the implications of missing data on time series of different granularities, such as minutely and hourly, and exploring a gamut of techniques that can be used to address the missing data issue (e.g., approximate the data using interpolation, regression, ensemble methods, etc.). Arun then walks you through how the techniques can be leveraged using real data.
Wolfram Alpha (also styled Wolfram Alpha and Wolfram Alpha) is a computational knowledge engine or answer engine developed by Wolfram Research. It is an online service that answers factual queries directly by computing the answer from externally sourced "curated data", rather than providing a list of documents or web pages that might contain the answer as a search engine might.
Nikhil Garg, Engineering Manager, Quora at MLconf SF 2016MLconf
Building a Machine Learning Platform at Quora: Each month, over 100 million people use Quora to share and grow their knowledge. Machine learning has played a critical role in enabling us to grow to this scale, with applications ranging from understanding content quality to identifying users’ interests and expertise. By investing in a reusable, extensible machine learning platform, our small team of ML engineers has been able to productionize dozens of different models and algorithms that power many features across Quora.
In this talk, I’ll discuss the core ideas behind our ML platform, as well as some of the specific systems, tools, and abstractions that have enabled us to scale our approach to machine learning.
Using OpenAI Gym and GNU Radio to Improve 5G
The openAI foundation built with Gym a toolkit for developing reinforcement learning algorithms and applying them in different environments.
GNU Radio on the other hand is a free open source tool for software radio and signal processing.
Those two tools combined create a powerful framework for researchers when it comes to applying machine learning approaches to radio-related problems.
In this talk, we will focus on how we can improve the next generation of mobile communication using reinforcement learning and open-source software, accessible for everyone.
A Fast Decision Rule Engine for Anomaly DetectionDatabricks
Description: We present a supervised anomaly detection approach that is scalable and interpretable. It works with tabular data and searches over all decision rules for the anomaly class involving one or two features. It creates a classifier out of all rules meeting user-specified precision and recall constraints, classifying a test example as an anomaly if any of the rules fire. Overlapping decision rules can be pruned to reduce model complexity, leaving a small number of simple rules that a user can easily understand. Our system operates on Pandas DataFrames and has a high-performance C++ backend with experimental GPU and FPGA acceleration available. It is available open-source at https://github.com/jjthomas/rule_engine
Brains, Data, and Machine Intelligence (2014 04 14 London Meetup)Numenta
Jeff will discuss the Brains, Data, Machine Intelligence, Cortical Learning Algorithm he developed and the Numenta Platform for Intelligent Computing (NuPIC).
A very distilled introduction to the concepts of Hierarchical Temporal Memory (HTM) and Sparse Distributed Representations (SDRs) as implemented by Numenta.
Exploration of U-Net and Support Vector Machine classification methods for UAV multispectral image segmentation
Recently, many solutions have been introduced to accurately and automatically analyze data acquired with Unmanned Aerial Vehicles (UAVs), in particular by relying on algorithms based on Artificial Intelligence (AI) techniques. Among these, the most popular are those belonging to the category of neural networks. These techniques allow the development of ad-hoc and end-to-end solutions for the classification and segmentation of different object categories through the analysis of high-resolution multispectral images. In our research, two main methodologies have been explored for the automatic segmentation of crop rows from multispectral images acquired with UAVs. The first is based on Support Vector Machines, know to handle well overfitting issues, and the other through the implementation of “U-Net”, a state-of-the-art Convolution Neural Network
Anomaly detection in real-time data streams using HeronArun Kejariwal
Twitter has become the de facto medium for consumption of news in real time, and billions of events are generated and analyzed on a daily basis. To analyze these events, Twitter designed its own next-generation streaming system, Heron. Arun Kejariwal and Karthik Ramasamy walk you through how Heron is used to detect anomalies in real-time data streams. Although there’s been over 75 years of prior work in anomaly detection, most of the techniques cannot be used off the shelf because they’re not suitable for high-velocity data streams. Arun and Karthik explain how to make trade-offs between accuracy and speed and discuss incremental approaches that marry sampling with robust measures such as median and MCD for anomaly detection.
Energy Monitoring With Self-taught Deep NetworkYiqun Hu
This is the presentation of my talk in O'Reilly Strata Data Conference Singapore 2017. It is about how we can extract useful knowledge from unlabelled time series to help energy monitoring applications.
Data Data Everywhere: Not An Insight to Take Action UponArun Kejariwal
The big data era is characterized by ever-increasing velocity and volume of data. Over the last two or three years, several talks at Velocity have explored how to analyze operations data at scale, focusing on anomaly detection, performance analysis, and capacity planning, to name a few topics. Knowledge sharing of the techniques for the aforementioned problems helps the community to build highly available, performant, and resilient systems.
A key aspect of operations data is that data may be missing—referred to as “holes”—in the time series. This may happen for a wide variety of reasons, including (but not limited to):
# Packets being dropped due to unresponsive downstream services
# A network hiccup
# Transient hardware or software failure
# An issue with the data collection service
“Holes” in the time series on data analysis can potentially skew the analysis of data. This in turn can materially impact decision making. Arun Kejariwal presents approaches for analyzing operations data in the presence of “holes” in the time series, highlighting how missing data impacts common data analysis such as anomaly detection and forecasting, discussing the implications of missing data on time series of different granularities, such as minutely and hourly, and exploring a gamut of techniques that can be used to address the missing data issue (e.g., approximate the data using interpolation, regression, ensemble methods, etc.). Arun then walks you through how the techniques can be leveraged using real data.
Wolfram Alpha (also styled Wolfram Alpha and Wolfram Alpha) is a computational knowledge engine or answer engine developed by Wolfram Research. It is an online service that answers factual queries directly by computing the answer from externally sourced "curated data", rather than providing a list of documents or web pages that might contain the answer as a search engine might.
Nikhil Garg, Engineering Manager, Quora at MLconf SF 2016MLconf
Building a Machine Learning Platform at Quora: Each month, over 100 million people use Quora to share and grow their knowledge. Machine learning has played a critical role in enabling us to grow to this scale, with applications ranging from understanding content quality to identifying users’ interests and expertise. By investing in a reusable, extensible machine learning platform, our small team of ML engineers has been able to productionize dozens of different models and algorithms that power many features across Quora.
In this talk, I’ll discuss the core ideas behind our ML platform, as well as some of the specific systems, tools, and abstractions that have enabled us to scale our approach to machine learning.
Using OpenAI Gym and GNU Radio to Improve 5G
The openAI foundation built with Gym a toolkit for developing reinforcement learning algorithms and applying them in different environments.
GNU Radio on the other hand is a free open source tool for software radio and signal processing.
Those two tools combined create a powerful framework for researchers when it comes to applying machine learning approaches to radio-related problems.
In this talk, we will focus on how we can improve the next generation of mobile communication using reinforcement learning and open-source software, accessible for everyone.
A Fast Decision Rule Engine for Anomaly DetectionDatabricks
Description: We present a supervised anomaly detection approach that is scalable and interpretable. It works with tabular data and searches over all decision rules for the anomaly class involving one or two features. It creates a classifier out of all rules meeting user-specified precision and recall constraints, classifying a test example as an anomaly if any of the rules fire. Overlapping decision rules can be pruned to reduce model complexity, leaving a small number of simple rules that a user can easily understand. Our system operates on Pandas DataFrames and has a high-performance C++ backend with experimental GPU and FPGA acceleration available. It is available open-source at https://github.com/jjthomas/rule_engine
Brains, Data, and Machine Intelligence (2014 04 14 London Meetup)Numenta
Jeff will discuss the Brains, Data, Machine Intelligence, Cortical Learning Algorithm he developed and the Numenta Platform for Intelligent Computing (NuPIC).
A very distilled introduction to the concepts of Hierarchical Temporal Memory (HTM) and Sparse Distributed Representations (SDRs) as implemented by Numenta.
"Kate, a Platform for Machine Intelligence" by Wayne Imaino, IBM Researchdiannepatricia
Wayne Imaino, Distinguished Research Staff Member at IBM Almaden Research Center, currently working to develop machine intelligence, made this presentation as part of the Cognitive Systems Institute Speaker Series on Jan 28, 2016.
Why Neurons have thousands of synapses? A model of sequence memory in the brainNumenta
Presentation given by Yuwei Cui, Numenta Research Engineer at Beijing Normal University. December 2015.
Collaborators: Jeff Hawkins, Subutai Ahmad, Chetan Surpur
SmartData Webinar: Applying Neocortical Research to Streaming AnalyticsDATAVERSITY
We are witnessing an explosion of sensors and machine generated data. Every server, every building, and every device generates a continuous stream of information that is ever changing and potentially valuable. The existing big data paradigm requires storing data for batch analysis, and extensive modeling by a human expert, prior to deployment. This is incredibly inefficient and cannot scale.
In this webinar, Ahmad will describe a new paradigm for streaming data algorithms, based on recent neuroscience findings and on the computational properties of the neocortex. These systems are highly automated, adapt to changing statistics, and naturally deal with temporal data streams. Many of the core ideas have been implemented in the open source project NuPIC, and validated in commercial anomaly detection and predictive maintenance applications. Given the massive increase in the number of data sources, a general-purpose automated approach is the only scalable way to effectively analyze and act on continuously streaming information.
Data Summer Conf 2018, “Architecting IoT system with Machine Learning (ENG)” ...Provectus
In this presentation, the speaker will share his experiences from building successful IoT systems. He will also explain why many IoT systems fail to get traction and how Machine Learning can help in that. Finally, he will talk about the right system architecture and touch upon some of the ML algorithms for IoT systems.
Machine learning, or predictive analytics have started entering into our daily life. Businesses and enterprises could use predictive analytics to improve efficiency, improve user experience, as well as to create new business opportunities. This talk will present WSO2 Machine Learner, our experiences of predicting Super Bowl winners, and few real life use cases. Furthermore, talk will discuss open challenges and problems people are working on.
Shiva Amiri, Chief Product Officer, RTDS Inc. at MLconf SEA - 5/01/15MLconf
Incorporating the Real Time Component into Analytics and Machine Learning: Many industries and organizations today want to harness the power of big data analytics and machine learning for its potential to improve margins, enhance discoveries, give insight into the business, and enable fast data driven decisions. The challenges include inability and/or difficulties in using available systems, not knowing where to start or which tools make sense for a particular problem, and dealing with data sets that are too big, too fast, or too complicated to handle with traditional systems.
RTDS Inc. has developed SymetryMLTM which are technologies for zero latency machine learning and analytics/exploration of very large datasets in real time, with a focus on speed, accuracy and simplicity. Our goal has been to cut the memory footprint required to learn large data sets, “reducer” functionality to automatically select the best attributes for model creation and build models on the fly. SymetryMLTM is also designed for easy integration into existing business processes via either an easy to use Web-UI or RESTful APIs.
This talk will explore some of the functionality of these systems including real time exploration of data, fast multi-variate model prototyping, and our use of GPUs and parallelization. An example of brain related data and the complexities of analytics will be discussed as well as a brief overview of other verticals we are exploring. Our work is geared towards making big data make sense in real time and enable users to gain insights faster than traditional methods.
Meetup sthlm - introduction to Machine Learning with demo casesZenodia Charpy
Data science and Machine Learning
Machine Learning vs Artificial Intelligence
Machine Learning Algorithms
How to choose ML algorithm mindmap
Supervised Learning generic flow
Unsupervised Learning generic flow
Example cases for supervised and unsupervised learning
Introduction of streaming data, difference between batch processing and stream processing, Research issues in streaming data processing, Performance evaluation metrics , tools for stream processing.
Data Science in the Real World: Making a Difference Srinath Perera
We use the terms “Big Data” and “Data Science” for use of data processing to make sense of the world around us. Spanning many fields, Big Data brings together technologies like Distributed Systems, Machine Learning, Statistics, and Internet of Things together. It is a multi-billion-dollar industry including use cases like targeted advertising, fraud detection, product recommendations, and market surveys. With new technologies like Internet of Things (IoT), these use cases are expanding to scenarios like Smart Cities, Smart health, and Smart Agriculture.
These usecases use basic analytics, advanced statistical methods, and predictive technologies like Machine Learning. However, it is not just about crunching the data. Some usecases like Urban Planning can be slow, and there is enough time to process the data. However, with use cases like traffic, patient monitoring, surveillance the the value of results degrades much faster with time and needs results within milliseconds to seconds. Collecting data from many sources, cleaning them up, processing them using computation clusters, and doing all these fast is a major challenge.
This talk will discuss motivation behind big data and data science and how it can make a difference. Then it will discuss the challenges, systems, and methodologies for implementing and sustaining a data science pipeline.
Big Data & Machine Learning - TDC2013 Sao PauloOCTO Technology
BigData and Machine Learning: Usage and Opportunities for your IT department
Talk presented at The Developer Conference in São Paulo - 12/0713
Mathieu DESPRIEE
Keynote presentation from ECBS conference. The talk is about how to use machine learning and AI in improving software engineering. Experiences from our project in Software Center (www.software-center.se).
Machine Learning open studio solution for data scientists & developersActiveeon
Machine Learning Open Studio (ML-OS) is an interactive graphical interface that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale. It provides a rich set of generic machine learning tasks that can be connected together to build either basic or complex machine learning workflows for various use cases such as: fraud detection, text analysis, online offer recommendations, prediction of equipment failures, facial expression analysis, etc. These tasks are open source and can be easily customized according to your needs. ML-OS can schedule and orchestrate executions while optimising the use of computational resources. Usage of resources (e.g. CPU, GPU, local, remote nodes) can be easily monitored.
Similar to Streaming Analytics: It's Not the Same Game (20)
Brains@Bay Meetup: A Primer on Neuromodulatory Systems - Srikanth RamaswamyNumenta
Meetup page: https://www.meetup.com/Brains-Bay/events/284481247/
Neuromodulators are signalling chemicals in the brain, which control the emergence of adaptive learning and behaviour. Neuromodulators including dopamine, acetylcholine, serotonin and noradrenaline operate on a spectrum of spatio-temporal scales in tandem and opposition to reconfigure functions of biological neural networks and to regulate global cognition and state transition. Although neuromodulators are important in shaping cognition, their phenomenology is yet to be fully realized in deep neural networks (DNNs). In this talk, we will give an overview of the biological organizing principles of neuromodulators in adaptive cognition and highlight the competition and cooperation across neuromodulators.
Brains@Bay Meetup: How to Evolve Your Own Lab Rat - Thomas MiconiNumenta
Meetup page: https://www.meetup.com/Brains-Bay/events/284481247/
A hallmark of intelligence is the ability to learn new flexible, cognitive behaviors - that is, behaviors that require discovering, storing and exploiting novel information for each new instance of the task. In meta-learning, agents are trained with external algorithms to learn one specific cognitive task. However, animals are able to pick up such cognitive tasks automatically, as a result of their evolved neural architecture and synaptic plasticity mechanisms, including neuromodulation. Here we evolve neural networks, endowed with plastic connections and reward-based neuromodulation, over a sizable set of simple meta-learning tasks based on a framework from computational neuroscience. The resulting evolved networks can automatically acquire a novel simple cognitive task, never seen during evolution, through the spontaneous operation of their evolved neural organization and plasticity system. We suggest that attending to the multiplicity of loops involved in natural learning may provide useful insight into the emergence of intelligent behavior.
Brains@Bay Meetup: The Increasing Role of Sensorimotor Experience in Artifici...Numenta
We receive information about the world through our sensors and influence the world through our effectors. Such low-level data has gradually come to play a greater role in AI during its 70-year history. I see this as occurring in four steps, two of which are mostly past and two of which are in progress or yet to come. The first step was to view AI as the design of agents which interact with the world and thereby have sensorimotor experience; this viewpoint became prominent in the 1980s and 1990s. The second step was to view the goal of intelligence in terms of experience, as in the reward signal of optimal control and reinforcement learning. The reward formulation of goals is now widely used but rarely loved. Many would prefer to express goals in non-experiential terms, such as reaching a destination or benefiting humanity, but settle for reward because, as an experiential signal, reward is directly available to the agent without human assistance or interpretation. This is the pattern that we see in all four steps. Initially a non-experiential approach seems more intuitive, is preferred and tried, but ultimately proves a limitation on scaling; the experiential approach is more suited to learning and scaling with computational resources. The third step in the increasing role of experience in AI concerns the agent’s representation of the world’s state. Classically, the state of the world is represented in objective terms external to the agent, such as “the grass is wet” and “the car is ten meters in front of me”, or with probability distributions over world states such as in POMDPs and other Bayesian approaches. Alternatively, the state of the world can be represented experientially in terms of summaries of past experience (e.g., the last four Atari video frames input to DQN) or predictions of future experience (e.g., successor representations). The fourth step is potentially the biggest: world knowledge. Classically, world knowledge has always been expressed in terms far from experience, and this has limited its ability to be learned and maintained. Today we are seeing more calls for knowledge to be predictive and grounded in experience. After reviewing the history and prospects of the four steps, I propose a minimal architecture for an intelligent agent that is entirely grounded in experience.
Brains@Bay Meetup: Open-ended Skill Acquisition in Humans and Machines: An Ev...Numenta
In this talk, I will propose a conceptual framework sketching a path toward open-ended skill acquisition through the coupling of environmental, morphological, sensorimotor, cognitive, developmental, social, cultural and evolutionary mechanisms. I will illustrate parts of this framework through computational experiments highlighting the key role of intrinsically motivated exploration in the generation of behavioral regularity and diversity. Firstly, I will show how some forms of language can self-organize out of generic exploration mechanisms without any functional pressure to communicate. Secondly, we will see how language — once invented — can be recruited as a cognitive tool that enables compositional imagination and bootstraps open-ended cultural innovation.
For more:
Brains@Bay Meetup: The Effect of Sensorimotor Learning on the Learned Represe...Numenta
Most current deep neural networks learn from a static data set without active interaction with the world. We take a look at how learning through a closed loop between action and perception affects the representations learned in a DNN. We demonstrate how these representations are significantly different from DNNs that learn supervised or unsupervised from a static dataset without interaction. These representations are much sparser and encode meaningful content in an efficient way. Even an agent who learned without any external supervision, purely through curious interaction with the world, acquires encodings of the high dimensional visual input that enable the agent to recognize objects using only a handful of labeled examples. Our results highlight the capabilities that emerge from letting DNNs learn more similar to biological brains, though sensorimotor interaction with the world.
For more:
SBMT 2021: Can Neuroscience Insights Transform AI? - Lawrence SpracklenNumenta
Numenta's Director of ML Architecture Lawrence Spracklen presented a talk at the SBMT Annual Congress on July 10th, 2021. He talked about how neuroscience principles can inspire better machine learning algorithms.
FPGA Conference 2021: Breaking the TOPS ceiling with sparse neural networks -...Numenta
Nick Ni (Xilinx) and Lawrence Spracklen (Numenta) presented a talk at the FGPA Conference Europe on July 8th, 2021. In this talk, they presented a neuroscience approach to optimize state-of-the-art deep learning networks into sparse topology and how it can unlock significant performance gains on FPGAs without major loss of accuracy. They then walked through the FPGA implementation where they exploited the advantage of sparse networks with a unique Domain Specific Architecture (DSA).
BAAI Conference 2021: The Thousand Brains Theory - A Roadmap for Creating Mac...Numenta
Jeff Hawkins presented a talk on "The Thousand Brains Theory: A Roadmap to Machine Intelligence" at the Beijing Academy of Artificial Intelligence Conference on 1st June 2021. In this talk, he discussed the key components of The Thousand Brains Theory and Numenta's recent work.
Jeff Hawkins NAISys 2020: How the Brain Uses Reference Frames, Why AI Needs t...Numenta
Jeff Hawkins presents a talk on "How the Brain Uses Reference Frames to Model the World, Why AI Needs to do the Same." In this talk, he gives an overview of The Thousand Brains Theory and discusses how machine intelligence can benefit from working on the same principles as the neocortex.
This talk was first presented at the NAISys conference on November 10, 2020. You can find a re-recording of the talk here: https://youtu.be/mGSG7I9VKDU
OpenAI’s GPT 3 Language Model - guest Steve OmohundroNumenta
In this research meeting, guest Stephen Omohundro gave a fascinating talk on GPT-3, the new massive OpenAI Natural Language Processing model. He reviewed the network architecture, training process, and results in the context of past work. There was extensive discussion on the implications for NLP and for Machine Intelligence / AGI.
Link to GPT-3 paper: https://arxiv.org/abs/2005.14165
Link to YouTube recording of Steve's talk: https://youtu.be/0ZVOmBp29E0
CVPR 2020 Workshop: Sparsity in the neocortex, and its implications for conti...Numenta
Numenta VP Research Subutai Ahmad presents a talk on "Sparsity in the Neocortex and its Implications for Continual Learning" at the virtual CVPR 2020 workshop. In this talk, he discusses how continuous learning systems can benefit from sparsity, active dendrites and other neocortical mechanisms.
The Thousand Brains Theory: A Framework for Understanding the Neocortex and B...Numenta
Recent advances in reverse engineering the neocortex reveal that it is a highly-distributed sensory-motor modeling system. Each cortical column learns complete models of observed objects through movement and sensation. The columns use long-range connections to vote on what objects are currently being observed. In this talk, we introduce the key elements of this theory and describe how these elements can be introduced into current machine learning techniques to improve their capabilities, robustness, and power requirements.
Jeff Hawkins Human Brain Project Summit Keynote: "Location, Location, Locatio...Numenta
Jeff Hawkins delivered this keynote presentation at the 2018 Human Brain Project Summit Open Day in Maastricht, the Netherlands on October 15, 2018. A screencast recording of the slides is also available at: https://numenta.com/resources/videos/jeff-hawkins-human-brain-project-screencast/
Location, Location, Location - A Framework for Intelligence and Cortical Comp...Numenta
Jeff Hawkins gave this presentation as part of the Johns Hopkins APL Colloquium Series on Septemer 21, 2018.
View the video of the talk here: https://numenta.com/resources/videos/jeff-hawkins-johns-hopkins-apl-talk/
Have We Missed Half of What the Neocortex Does? A New Predictive Framework ...Numenta
Numenta VP of Research Subutai Ahmad delivered this presentation at the Centre for Theoretical Neuroscience, University of Waterloo on October 2, 2018.
The Biological Path Toward Strong AI by Matt Taylor (05/17/18)Numenta
These are Matt Taylor's slides from the AI Singapore Meetup on May 17, 2018.
Abstract:
Today’s wave of AI technology is still being driven by the ANN neuron pioneered decades ago. Hierarchical Temporal Memory (HTM) is a realistic biologically-constrained model of the pyramidal neuron reflecting today’s most recent neocortical research. This talk will describe and visualize core HTM concepts like sparse distributed representations, spatial pooling and temporal memory. Strong AI is a common goal of many computer scientists. So far, machine learning techniques have created amazing results in narrow fields, but haven’t produced something we could all call “intelligent”. Given recent advances in neuroscience research, we know a lot more about how neurons work together now than we did when ANNs were created. We believe systems with a more realistic neuronal model will be more likely to produce Strong AI. Hierarchical Temporal Memory is a theory of intelligence based upon neuroscience research. The neocortex is the seat of intelligence in the brain, and it is structurally homogeneous throughout. This means a common algorithm is processing all your sensory input, no matter which sense. We believe we have discovered some of the foundational algorithms of the neocortex, and we’ve implemented them in software. I’ll show you how they work with detailed dynamic visualizations of Sparse Distributed Representations, Spatial Pooling, and Temporal Memory.
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...John Andrews
SlideShare Description for "Chatty Kathy - UNC Bootcamp Final Project Presentation"
Title: Chatty Kathy: Enhancing Physical Activity Among Older Adults
Description:
Discover how Chatty Kathy, an innovative project developed at the UNC Bootcamp, aims to tackle the challenge of low physical activity among older adults. Our AI-driven solution uses peer interaction to boost and sustain exercise levels, significantly improving health outcomes. This presentation covers our problem statement, the rationale behind Chatty Kathy, synthetic data and persona creation, model performance metrics, a visual demonstration of the project, and potential future developments. Join us for an insightful Q&A session to explore the potential of this groundbreaking project.
Project Team: Jay Requarth, Jana Avery, John Andrews, Dr. Dick Davis II, Nee Buntoum, Nam Yeongjin & Mat Nicholas
Explore our comprehensive data analysis project presentation on predicting product ad campaign performance. Learn how data-driven insights can optimize your marketing strategies and enhance campaign effectiveness. Perfect for professionals and students looking to understand the power of data analysis in advertising. for more details visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/
Adjusting primitives for graph : SHORT REPORT / NOTESSubhajit Sahu
Graph algorithms, like PageRank Compressed Sparse Row (CSR) is an adjacency-list based graph representation that is
Multiply with different modes (map)
1. Performance of sequential execution based vs OpenMP based vector multiply.
2. Comparing various launch configs for CUDA based vector multiply.
Sum with different storage types (reduce)
1. Performance of vector element sum using float vs bfloat16 as the storage type.
Sum with different modes (reduce)
1. Performance of sequential execution based vs OpenMP based vector element sum.
2. Performance of memcpy vs in-place based CUDA based vector element sum.
3. Comparing various launch configs for CUDA based vector element sum (memcpy).
4. Comparing various launch configs for CUDA based vector element sum (in-place).
Sum with in-place strategies of CUDA mode (reduce)
1. Comparing various launch configs for CUDA based vector element sum (in-place).
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Subhajit Sahu
Abstract — Levelwise PageRank is an alternative method of PageRank computation which decomposes the input graph into a directed acyclic block-graph of strongly connected components, and processes them in topological order, one level at a time. This enables calculation for ranks in a distributed fashion without per-iteration communication, unlike the standard method where all vertices are processed in each iteration. It however comes with a precondition of the absence of dead ends in the input graph. Here, the native non-distributed performance of Levelwise PageRank was compared against Monolithic PageRank on a CPU as well as a GPU. To ensure a fair comparison, Monolithic PageRank was also performed on a graph where vertices were split by components. Results indicate that Levelwise PageRank is about as fast as Monolithic PageRank on the CPU, but quite a bit slower on the GPU. Slowdown on the GPU is likely caused by a large submission of small workloads, and expected to be non-issue when the computation is performed on massive graphs.
1. Strata + Hadoop World, 2015
February 20, 2015
Subutai Ahmad
sahmad@numenta.com
Streaming Analytics:
It’s Not The Same Game
2. Objectives for next generation:
Generate predictions every 15-minutes
Track all product categories and geographies (hundreds of
thousands)
React rapidly to changes
Problems:
Cumbersome data infrastructure
Algorithm approach completely unclear
Slow business processes
Revenue Forecasting Customer Story
10pm
Team of 10 analysts
“Dear CEO, today’s revenue forecast is $63.4M.”
3. 2. Look at data 3. Build models
Problem: Doesn’t scale with data velocity and number of models
1. Store data
Streaming data
Automated model creation
Continuous learning
Temporal inference
Predictions
Anomalies
Actions
Past
Future
Data: Past and Future
Solution: Streaming data infrastructure
New algorithm approach
Optimized business processes
4. Talk Outline
1) Challenges for traditional machine learning algorithms
2) A new approach to streaming, based on neuroscience
3) Streaming applications
12. Functional Properties Of The Neocortex
1) Hierarchy of nearly identical regions
- common algorithm
retina
cochlea
somatic
2) Regions are mostly sequence memory
- inference
- motor
data stream
3) Sparse Distribution Representations
- common data structure
motor control
4) Every region is continually learning
- Fully unsupervised
“Hierarchical Temporal Memory” (HTM)
13. Physical Architecture of the Cortex
Cortical region Layers with
columns
Neurons with
thousands of
synapses
Dendrites act as
coincidence detectors
HTM Learning Algorithms
14. HTM Example
Time of DayEncoders Sensor Value
Data
Spatial Pooler
Temporal Memory
HTM
Learning
Algorithms
Predictions
Anomalies
Models common spatial patterns and
temporal sequences in the stream
At every time step improve representation of
that spatial pattern and that transition
At each time step Temporal Memory
makes multiple predictions about
what might occur next
SDR
15. HTM Learning Algorithm Codebase
Models a single layer of cortex
1) High capacity memory-based system
2) Models complex high-order temporal sequences
3) Makes predictions and detects anomalies
4) Continuously learning
5) No sensitive parameters
Basic building block of neocortex/Machine Intelligence
Whitepaper and full source code available
at numenta.org & github.com/numenta
16. HTM
Encoder
SDR
Prediction
Point anomaly
Time average
Historical comparison
Anomaly score
Metric(s)
System
Anomaly
Scores
&
Predictions
HTM Engine For Streaming Analytics
HTM
Encoder
SDR
Prediction
Point anomaly
Time average
Historical comparison
Anomaly score
SDR
Metric N
.
.
.
18. Grok: Anomaly Detection For Amazon Web Services
Unique value of HTM algorithms
Automated model creation: configure hundreds of models in minutes
Continuously learning: automatically adapts to changes
Detects sophisticated temporal anomalies
19. 3) Anomaly Detection in Geospatial Tracking Data
Fleets, Planes, Materiel, Kids, Pets
CLA
Encoder
SDRs
Prediction
Anomaly Detection
Classification
GPS+ Velocity
Trick: convert GPS coordinates into an SDR
- Represents both location and speed
- Works anywhere on Earth or in space
After input is encoded as an SDR, learning algorithm is agnostic
- This process needs to be done once per sensor type
25. These HTM Applications Use Exact Same Code Base
HTM learning algorithms
Identical learning parameters
Suitable for many data types
GROK
Server anomalies
Rogue human
behavior
Geospatial
tracking
Stock
anomalies
Social media
streams (Twitter)
26. Future of Data is Streaming Data
- High velocity data streams that change often
- Massive number of models
- Existing batch paradigm cannot scale
Streaming Analytics & Algorithms
- Must move towards:
Automated model building
Continuous learning
Temporal modeling
- We can learn how to do this from neuroscience
- HTM learning algorithms demonstrate that this is possible
Summary(Numenta Platform for Intelligent Computing)
27. Learn More
- Documentation, videos at numenta.com/learn
- Source code and examples at github.com/numenta
Get Involved
- Provide feedback to info@numenta.com
- Participate in NuPIC
- Active mailing lists
- Try out Grok, on AWS Marketplace