7 steps for highly effective deep neural networks

•

0 likes•501 views

Natalino will provide a fast-paced lecture on recognizing handwritten numbers with ANNs. In the past years much progress has been booked on the recognition of images: multi-layer perceptrons, inception networks, skip and residual nets, lstm, attention networks. In this talk, Natalino will go through 7 different architectures on how to classify handwritten numbers using keras and tensorflow. For each neural network architecture, he will spend some time on the theory, the coding, and the quality of the results. Code and Fun included!. original file: https://drive.google.com/file/d/0BwNrPuGaMi8PbVhUYUVKWUhGRjQ

Data & Analytics

7 steps for highly effective
deep neural networks
Natalino Busa - Head of Data Science

2 Natalino Busa - @natbusa
O’Reilly Author and Speaker
Teradata EMEA Practice Lead on Open Source Technologies
Teradata Principal Data Scientist
ING Group Enterprise Architect: Cybersecurity, Marketing, Fintech
Cognitive Finance Group Advisory Board Member
Philips Senior Researcher, Data Architect
Linkedin and Twitter:
@natbusa

3
Natalino Busa - @natbusa
X: independent variable
Y: dependent variable
Linear Regression:
How to best fit a line to some data

4
Natalino Busa - @natbusa
input : 784 numbers
output: 10 classes
28x28 pixels
Classification: Handwritten digits

5
Natalino Busa - @natbusa
Sharing the (Not so) Secret Lore:
Keras & Tensorflow
Some smaller projects:
Tflearn, Tensorlayers
http://keras.io/

6
Natalino Busa - @natbusa
1: Single Layer Perceptron
“dendrites”
Axon’s
response
Activation function

7
Natalino Busa - @natbusa
More activation functions:
https://en.wikipedia.org/wiki/Activation_function

8
Natalino Busa - @natbusa
Single Layer Neural Network
Takes: n-input features: Map them to a soft “binary” space
∑
x1
x2
xn
f
1: Single Layer Perceptron (binary classifier)

9
Natalino Busa - @natbusa
1: Single Layer Perceptron (multi-class classifier)
Values between 0 and 1
Sum of all outcomes = 1
It produces an estimate of a probability!
From soft binary space to predicting probabilities:
Take n inputs, Divide by the sum of the predicted values
∑
x1
x2
xn
f
∑ f
softmax ‘1’: 95%
‘8’: 5%

10
Natalino Busa - @natbusa
1: Single Layer Perceptron
Minimize costs:
The cost function depends on:
- Parameters of the model
- How the model “composes”
Goal :
Reduce the mean probability error
modify the parameters to reduce the error! Vintage math from last century

11
Natalino Busa - @natbusa
Supervised Learning
Stack layers of perceptrons
- Feed Forward Network
- Scoring goes
from input to output
- Back propagate the error
from output to input
SOFTMAX
Input parameters
classes (estimated probabilities)
Feed-forwardfunctions
Cost function
supervised : actual output
BackPropagateErrors

13
Natalino Busa - @natbusa
1. Single Layer Perceptron

14
Natalino Busa - @natbusa
1. Single Layer Perceptron

15
Natalino Busa - @natbusa
1. Single Layer Perceptron

16
Natalino Busa - @natbusa
1. Single Layer Perceptron

20
Natalino Busa - @natbusa
2. Multi Layer Perceptron

21
Natalino Busa - @natbusa
2. Multi Layer Perceptron

22
Natalino Busa - @natbusa
2. Multi Layer Perceptron

23
Natalino Busa - @natbusa
2. Multi Layer Perceptron

24
Natalino Busa - @natbusa
3. Convolution
From Krizehvsky et al. (2012)

25
Natalino Busa - @natbusa
3. Convolution
diagrams:
By Aphex34 - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=45659236
By Aphex34 - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=45673581
CC0, https://en.wikipedia.org/w/index.php?curid=48817276
convolution Max pooling RELU / ELU

26
Natalino Busa - @natbusa
3. Convolution

27
Natalino Busa - @natbusa
3. Convolution

28
Natalino Busa - @natbusa
3. Batch Normalization

29
Natalino Busa - @natbusa
3. Batch Normalization ( example for MLP)

30
Natalino Busa - @natbusa
3. Batch Normalization ( example for MLP)

31
Natalino Busa - @natbusa
3. Batch Normalization ( example for MLP)
Activation function:
SIGMOID

32
Natalino Busa - @natbusa
3. Batch Normalization ( example for MLP)
Activation function:
RELU

33
Natalino Busa - @natbusa
4. Regularization: Prevent overfitting in ANNs
- Batch Normalization
- RELU/ELU
- RESIDUAL / SKIP Networks
- DROP LAYER
- REDUCE PRECISION (HUFFMAN ENCODING)
In general ANN are parameters rich, constraining the
parameter space usually produces better results and speed
up the learning

34
Natalino Busa - @natbusa
5. Inception architectures
Cannot be stacked!

35
Natalino Busa - @natbusa
5. Inception architectures

36
Natalino Busa - @natbusa
5. Inception architectures
compression
Avoid
dimensions
explosion

37
Natalino Busa - @natbusa
5. Inception architectures

38
Natalino Busa - @natbusa
5. Inception architectures (top architecture)

39
Natalino Busa - @natbusa
6. Residual Networks
https://culurciello.github.io/tech/2016/06/04/nets.html

40
Natalino Busa - @natbusa
6. Residual Networks

41
Natalino Busa - @natbusa
6. Residual + Inception Networks

42
Natalino Busa - @natbusa
7. LSTM on Images

43
Natalino Busa - @natbusa
7. LSTM on Images

44
Natalino Busa - @natbusa
7. LSTM on Images
http://colah.github.io/posts/2015-08-Understanding-LSTMs/

45
Natalino Busa - @natbusa
7. LSTM on Images

46
Natalino Busa - @natbusa
7. LSTM on Images

47
Natalino Busa - @natbusa
7. LSTM on ConvNets (bonus slide)

48
Natalino Busa - @natbusa
All the codez :)
https://github.com/natbusa/deepnumbers

49
Natalino Busa - @natbusa
Meta- References
… just a few articles, but extremely dense in content.
A must read!
https://keras.io/
http://karpathy.github.io/neuralnets/
https://culurciello.github.io/tech/2016/06/04/nets.html
http://colah.github.io/posts/2015-08-Understanding-LSTMs/
https://gab41.lab41.org/batch-normalization-what-the-hey-d480039a9e3b

Jupyter notebooks are transforming the way we look at computing, coding, and science. But is this the only "data scientist experience" that this technology can provide? In this presentation we will look at how to create interactive web applications for data exploration and machine learning. In the background this code is still powered by the well-understood and well-documented Jupyter Notebooks. Code on github: https://github.com/natbusa/kernelgateway_demos

Data science apps: beyond notebooks

Natalino Busa

Jupyter notebooks are transforming the way we look at computing, coding and problem solving. But is this the only “data scientist experience” that this technology can provide? In this webinar, Natalino will sketch how you could use Jupyter to create interactive and compelling data science web applications and provide new ways of data exploration and analysis. In the background, these apps are still powered by well understood and documented Jupyter notebooks. They will present an architecture which is composed of four parts: a jupyter server-only gateway, a Scala/Spark Jupyter kernel, a Spark cluster and a angular/bootstrap web application.

Python for Data Science

Harri Hämäläinen

Python for data science

Tanzeel Ahmad Mujahid

Power of Python with Big Data

Edureka!

Programmers love Python because of how fast and easy it is to use. Python cuts development time in half with its simple to read syntax and easy compilation feature. Debugging your programs is a breeze in Python with its built in debugger. Python is continued to be a favourite option for data scientists who use it for building and using Machine learning applications and other scientific computations. Python has evolved as the most preferred Language for Data Analytics and the increasing search trends on python also indicates that Python is the next "Big Thing" and a must for Professionals in the Data Analytics domain.

Analyzing Data With Python

Sarah Guido

Sarah Guido gave a presentation on analyzing data with Python. She discussed several Python tools for preprocessing, analysis, and visualization including Pandas for data wrangling, scikit-learn for machine learning, NLTK for natural language processing, MRjob for processing large datasets in parallel, and ggplot for visualization. For each tool, she provided examples and use cases. She emphasized that the best tools depend on the type of data and analysis needs.

Python for Big Data Analytics

Edureka!

This document provides an overview and objectives of a Python course for big data analytics. It discusses why Python is well-suited for big data tasks due to its libraries like PyDoop and SciPy. The course includes demonstrations of web scraping using Beautiful Soup, collecting tweets using APIs, and running word count on Hadoop using Pydoop. It also discusses how Python supports key aspects of data science like accessing, analyzing, and visualizing large datasets.

Are we reaching a Data Science Singularity? How Cognitive Computing is emergi...

Big Data Spain

This document discusses tools for making NumPy and Pandas code faster and able to run in parallel. It introduces the Dask library, which allows users to work with large datasets in a familiar Pandas/NumPy style through parallel computing. Dask implements parallel DataFrames, Arrays, and other collections that mimic their Pandas/NumPy counterparts. It can scale computations across multiple cores on a single machine or across many machines in a cluster. The document provides examples of using Dask to analyze large CSV and text data in parallel through DataFrames and Bags. It also discusses scaling computations from a single laptop to large clusters.

Scientific Python

Eueung Mulyana

This document provides an overview and introduction to key Python packages for scientific computing and data science. It discusses Jupyter notebooks for interactive coding and visualization, NumPy for N-dimensional arrays and math operations, SciPy for scientific computing functions, matplotlib for plotting, and pandas for working with labeled data structures. The document emphasizes that NumPy provides foundational N-dimensional arrays, SciPy builds on this with additional mathematical and scientific routines, and matplotlib and pandas complement these with visualization and labeled data functionality.

Puppet 3: Present and Future Tense

Puppet

This document provides an overview and summary of Puppet 3 features from a presentation given by Eric Sorenson of Puppet Labs. Some key points include: - Puppet 3.1 focuses on performance improvements and platform support for systems like Solaris and Windows. - New features include pluginsync, lazy loading of plugins, and improved packaging support. - Hiera data is now integrated directly into Puppet's code via data bindings. - The language adds new features like unless and improved array handling. - Dynamic variable scoping has been removed, requiring explicit qualification of variables.

Puppet 3: Present and Future Tense

Eric Sorenson

This document provides an overview and summary of Puppet 3 features from a presentation given by Eric Sorenson of Puppet Labs. Some key points include: - Puppet 3.1 focuses on performance improvements and platform support for systems like Solaris and Windows. - New features include pluginsync, lazy loading of modules, and improved packaging support. - Hiera data is now integrated directly into Puppet's language via data bindings. - The document outlines plans to improve modularity, asynchronicity, and integration in future Puppet releases.

Understanding Jupyter notebooks using bioinformatics examples

Lynn Langit

Deep Learning and Tensorflow Implementation(딥러닝, 텐서플로우, 파이썬, CNN)_Myungyon Ki...

Myungyon Kim

Deep learning and Tensorflow implementation 2016.11.16 <Cotents> Feature Engineering Deep Neural Network Tensorflow Tensorflow Implementation Future works References This slides deals with several things about deep learning. ex) History of Deep learning, Several difficulties and breakthroughs. Things related to deep learning such as activation functions, perceptrons, Backpropagation, pre-train, drop-out, Convolutional Neural Network (CNN), Simple implementation of Tensor Flow, Python, and so on. 딥러닝, 기계학습, 머신러닝, 텐서플로우, 파이썬

Real-Time Anomoly Detection with Spark MLib, Akka and Cassandra by Natalino Busa

Spark Summit

This document discusses real-time anomaly detection using Spark MLlib, Akka and Cassandra. It describes using clustering algorithms like k-means and DBSCAN on streaming data to identify anomalous patterns in venues' visitor patterns and users' check-in locations. The Akka-Cassandra-Spark stack is proposed to enable fast writes to Cassandra, distributed and scalable computing using Spark, and real-time processing of streaming data using Akka actors and pipelines.

2951085 dzone-2016guidetobigdata

balu kvm

Developers are increasingly working with large datasets and high data processing demands. 22% work with datasets over 1TB in size, and 27% must process over 1000 messages per second. Hadoop remains the most commonly adopted big data technology, with 19% of developers planning to adopt it, but Spark is gaining ground at 14% adoption. Developers are generally familiar with machine learning algorithms like neural networks, but have less practical experience applying them. As data volumes and processing speeds continue growing, more developers will need to leverage distributed computing frameworks to efficiently handle their data.

Introduction to Neural Networks in Tensorflow

Nicholas McClure

Nick McClure gave an introduction to neural networks using Tensorflow. He explained the basic unit of neural networks as operational gates and how multiple gates can be combined. He discussed loss functions, learning rates, and activation functions. McClure also covered convolutional neural networks, recurrent neural networks, and applications such as image captioning and style transfer. He concluded by discussing resources for staying up to date with advances in machine learning.

Python coroutine

경섭 심

This document provides an overview of coroutines in Python. It defines coroutines as components that allow non-preemptive multitasking and multiple entry points through suspending and resuming execution. Coroutines are compared to subroutines and generators. Generators in Python are described as frozen stack frames that yield values. Decorators are introduced as callables that modify function definitions. The document discusses how coroutines address issues with threads in Python by providing lightweight concurrency with less memory usage and startup costs compared to threads.

Espresso: LinkedIn's Distributed Data Serving Platform (Talk)

Amy W. Tang

Using Deep Learning in Production Pipelines to Predict Consumers’ Interest wi...

Databricks

To optimize customer conversion in e-commerce and promote the right message to the right person at the right time it’s necessary to build powerful predictive models, which usually involves a lot of feature engineering to aggregate consumer related past events (clicks, page views, purchasesâ¦) and extract relevant signals. Recurrent Neural Networks (RNN) is a class of deep-learning techniques, especially good at working with sequences of inputs and at learning time-dependant patterns. They can learn from past customer behavior, and their internal state can then be used as latent features for downstream models. In this session, we will see how to use RNN algorithms like LSTM to learn from sequences of events, and produce new features to be used in predictive models. Come see how we managed to use those RNN networks in our Spark pipelines!

Large scale social networks analysis joclad 2013

Rui Sarmento

Machine Learning with Spark

elephantscale

Startup.Ml: Using neon for NLP and Localization Applications

Intel Nervana

This document provides an overview of developing deep learning models with the neon deep learning framework. It introduces deep learning concepts and the Nervana platform, then describes hands-on exercises for building models including a sentiment analysis model using LSTMs on an IMDB dataset. Key aspects of neon like model architecture, initialization, datasets, backends, and training are demonstrated. Finally, a demo is shown for training and inference of the sentiment analysis model.

A Platform for Accelerating Machine Learning Applications

NVIDIA Taiwan

Robert Sheen from HPE gave a presentation on machine learning applications and accelerating deep learning. He provided a quick introduction to neural networks, discussing their structure and how they are inspired by biological neurons. Deep learning requires high performance computing due to its computational intensity during training. Popular deep learning frameworks like CogX were also discussed, which provide tools and libraries to help build and optimize neural networks. Finally, several enterprise use cases for machine learning and deep learning were highlighted, such as in finance, healthcare, security, and geospatial applications.

ドキュメントを作りたくなってしまう魔法のツールSphinx

Takayuki Shimizukawa

11_Saloni Malhotra_SummerTraining_PPT.pptx

SaloniMalhotra23

1) The document describes a deep learning project to detect brain tumours using MRI scan images. Convolutional neural network models are developed and trained on a dataset containing MRI scans labeled as either normal or tumour. 2) A basic CNN model is built first with convolutional, pooling and flatten layers, achieving an accuracy of 78%. Then a more complex VGG16 CNN architecture model is built, achieving a higher accuracy of 92.3% for tumour detection. 3) The project aims to accurately analyze MRI scans to detect brain tumours using deep learning algorithms like CNNs, which can help improve diagnostics and life expectancy for patients.

Basic of python for data analysis

Pramod Toraskar

Data Production Pipelines: Legacy, practices, and innovation

Natalino Busa

Modern engineering requires machine learning engineers, who are needed to monitor and implement ETL and machine learning models in production. Natalino Busa shares technologies, techniques, and blueprints on how to robustly and reliably manage data science and ETL flows from inception to production. In particular, Natalino explains how to solve one of the most annoying problems in modern data pipelines—migrating and managing legacy ETL—by generating Spark jobs from a textual representation (NLP and SQL). Natalino also demonstrates an open source web UI implemented in React that transforms high-level representations to Spark code and shows how users are able to capture and discover data in the organization by accessing a metadata service. Natalino also introduces the datalabframework, a Jupyter-powered lightweight framework that allows machine learning scientists and engineers to build a robust production ML system only using notebooks. https://conferences.oreilly.com/strata/strata-sg/public/schedule/detail/64481

[Ai in finance] AI in regulatory compliance, risk management, and auditing

Natalino Busa

Similar to 7 steps for highly effective deep neural networks

Scaling PyData Up and Out

Travis Oliphant

Fast and Scalable Python

Travis Oliphant

Scientific Python

Eueung Mulyana

Puppet 3: Present and Future Tense

Puppet

Puppet 3: Present and Future Tense

Eric Sorenson

This document provides an overview and summary of Puppet 3 features from a presentation given by Eric Sorenson of Puppet Labs. Some key points include: - Puppet 3.1 focuses on performance improvements and platform support for systems like Solaris and Windows. - New features include pluginsync, lazy loading of modules, and improved packaging support. - Hiera data is now integrated directly into Puppet's language via data bindings. - The document outlines plans to improve modularity, asynchronicity, and integration in future Puppet releases.

Understanding Jupyter notebooks using bioinformatics examples

Lynn Langit

Deep Learning and Tensorflow Implementation(딥러닝, 텐서플로우, 파이썬, CNN)_Myungyon Ki...

Myungyon Kim

Real-Time Anomoly Detection with Spark MLib, Akka and Cassandra by Natalino Busa

Spark Summit

2951085 dzone-2016guidetobigdata

balu kvm

Introduction to Neural Networks in Tensorflow

Nicholas McClure

Python coroutine

경섭 심

Espresso: LinkedIn's Distributed Data Serving Platform (Talk)

Amy W. Tang

Using Deep Learning in Production Pipelines to Predict Consumers’ Interest wi...

Databricks

Large scale social networks analysis joclad 2013

Rui Sarmento

Machine Learning with Spark

elephantscale

Startup.Ml: Using neon for NLP and Localization Applications

Intel Nervana

A Platform for Accelerating Machine Learning Applications

NVIDIA Taiwan

ドキュメントを作りたくなってしまう魔法のツールSphinx

Takayuki Shimizukawa

11_Saloni Malhotra_SummerTraining_PPT.pptx

SaloniMalhotra23

Basic of python for data analysis

Pramod Toraskar

Similar to 7 steps for highly effective deep neural networks (20)

Scaling PyData Up and Out

Fast and Scalable Python

Scientific Python

Puppet 3: Present and Future Tense

Understanding Jupyter notebooks using bioinformatics examples

Deep Learning and Tensorflow Implementation(딥러닝, 텐서플로우, 파이썬, CNN)_Myungyon Ki...

Real-Time Anomoly Detection with Spark MLib, Akka and Cassandra by Natalino Busa

2951085 dzone-2016guidetobigdata

Introduction to Neural Networks in Tensorflow

Python coroutine

Espresso: LinkedIn's Distributed Data Serving Platform (Talk)

Using Deep Learning in Production Pipelines to Predict Consumers’ Interest wi...

Large scale social networks analysis joclad 2013

Machine Learning with Spark

Startup.Ml: Using neon for NLP and Localization Applications

A Platform for Accelerating Machine Learning Applications

ドキュメントを作りたくなってしまう魔法のツールSphinx

11_Saloni Malhotra_SummerTraining_PPT.pptx

Basic of python for data analysis

More from Natalino Busa

Data Production Pipelines: Legacy, practices, and innovation

Natalino Busa

[Ai in finance] AI in regulatory compliance, risk management, and auditing

Natalino Busa

Strata London 16: sightseeing, venues, and friends

Natalino Busa

Data in Action

Natalino Busa

Real-Time Anomaly Detection with Spark MLlib, Akka and Cassandra

Natalino Busa

We present a solution for streaming anomaly detection, named “Coral”, based on Spark, Akka and Cassandra. In the system presented, we run Spark to run the data analytics pipeline for anomaly detection. By running Spark on the latest events and data, we make sure that the model is always up-to-date and that the amount of false positives is kept low, even under changing trends and conditions. Our machine learning pipeline uses Spark decision tree ensembles and k-means clustering. Once the model is trained by Spark, the model’s parameters are pushed to the Streaming Event Processing Layer, implemented in Akka. The Akka layer will then score 1000s of event per seconds according to the last model provided by Spark. Spark and Akka communicate which each other using Cassandra as a low-latency data store. By doing so, we make sure that every element of this solution is resilient and distributed. Spark performs micro-batches to keep the model up-to-date while Akka detects the new anomalies by using the latest Spark-generated data model. The project is currently hosted on Github. Have a look at : http://coral-streaming.github.io

The evolution of data analytics

Natalino Busa

In the past decade a number of technologies have revolutionized the way we do analytics in banking. In this talk we would like to summarize this journey from classical statistical offline modeling to the latest real-time streaming predictive analytical techniques. In particular, we will look at hadoop and how this distributing computing paradigm has evolved with the advent of in-memory computing. We will introduce Spark, an engine for large-scale data processing optimized for in-memory computing. Finally, we will describe how to make data science actionable and how to overcome some of the limitations of current batch processing with streaming analytics.

Towards Real-Time banking API's: Introducing Coral, a web api for realtime st...

Natalino Busa

Banks are innovating. The purpose of this innovation is to transform bank services into meaningful and frictionless customer experiences. A key element in order to achieve that ambitious goal is by providing well tailored and reactive APIs and provide them as the building blocks for greater and smoother customer journeys and experiences. For these API’s to work, internal processes have to evolve as well from batch processing to real time event processing. In this talk, after providing a brief introduction of the streaming computing landscape, we describe a RESTful API called “Coral” meant to design and deploy customized and flexible data flows. The user can compose data flow for a number of data streaming goals such as on-the-fly data clustering and classifiers, streaming analytics, per-event predictive analysis , real time recommenders. Once the events are processed, Coral passes the resulting analysis as auctionable events for alerting, messaging or further processing to other systems. Coral is a flexible and generic event processing platform to transform streaming data into actionable events via a RESTful API. Those data flows are defined via the Web API by connecting together basic streaming processing elements named “coral actors”. The Coral framework manages those coral actors on a distributed and scalable architecture. Streaming and real time data processing and analytics are the key elements to an improved customer experience. In this way, you can get the most targeted processing for your domain (marketing customization, personalized recommenders, fraud detection, real time security alerting, etc.). This streaming “data flow” model implies processing customers’ events as soon as they enter via web APIs. This approach borrows a lot from distributed “data flow” concepts developed for processor architectures back in the 80’s. The “Coral” streaming processing engine is generic and built on top of world class libraries such as Akka and Spark, and fully exposed via a RESTful web API.

Streaming Api Design with Akka, Scala and Spray

Natalino Busa

The document proposes designing a streaming API that aggregates data from wristbands worn by runners. It defines endpoints to create a stream, add filters to the stream to calculate metrics like maximum heart rate per second within groups, ingest data, and retrieve the filtered output. The API is built on Akka actors for asynchronous and reactive processing and on Spray for building HTTP services on top of Akka actors.

Hadoop + Cassandra: Fast queries on data lakes, and wikipedia search tutorial.

Natalino Busa

Today’s services rely on massive amount of data to be processed, but require at the same time to be fast and responsive. Building fast services on big data batch- oriented frameworks is definitely a challenge. At ING, we have worked on a stack that can alleviate this problem. Namely, we materialize data model by map-reducing Hadoop queries from Hive to Cassandra. Instead of sinking the results back to hdfs, we propagate the results into Cassandra key-values tables. Those Cassandra tables are finally exposed via a http API front-end service.

Big data solutions for advanced marketing analytics

Natalino Busa

Our retail banking market demands now more than ever to stay close to our customers, and to carefully understand what services, products, and wishes are relevant for each customer at any given time. This sort of marketing research is often beyond the capacity of traditional BI reporting frameworks. In this talk, we illustrate how we team up data scientists and big data engineers in order to create and scale distributed analyses on a big data platform. By using Hadoop and open source statistical language and tools such R and Python, we can execute a variety of machine learning algorithms, and scale them out on a distributed computing framework.

Awesome Banking API's

Natalino Busa

How do you combine comprehensive analysis running on large amount of data with the demand for responsiveness of today's api services? This talk illustrates one of recipes that we currently use at ING to tackle this problem. Our analytical stack combines machine learning algorithms running on hadoop cluster and api services executed by an akka cluster. Cassandra is used as a 'latency adapter' between the fast and the slow path. Our api services are executed by the akka/spray layer. Those services consume both live data sources as well as intermediate results as promoted by the hadoop layer via cassandra. This approach allows us to provide internal api services which are both complete and responsive.

Yo. big data. understanding data science in the era of big data.

Natalino Busa

We talk a lot these days about data science, and how it will pave our paths with beautiful insights and unexpected new relations and connections in our given datasets, and even across datasets. But how to maintain the "Science" part in "Data Science"? After some time working in this field I appreciate more and more the critical thinking which has characterized the progress in science. Hypothesis, facts, prove and/or disprove the thesis. This is how science has progressed in the past centuries. This method has been formalized by Popper and categorize as non-science all disciplines where the statements cannot be falsified. In other words, if a statement cannot be disproved, we cannot talk of science, since there is no mechanism to left to verify the solution or to prove it wrong. When that happens the argument can still be accepted, but not scientifically accepted. Ways of accepting or refuting a non falsifiable statement are for instance based on aesthetic, authority or pragmatic or philosophical considerations. All valid but not scientific. This applies for instance to statements in the disciplines of politics, teology, ethics, etc. Science has definitely progressed since then. For instance, Bayesian networks and statistical inductions are currently part of the arsenal of the (data) scientist weapons. But, no matter how the baseline is set, critical thinking and a rigorous method are definitely helpful in understanding the results produced by science in particular when this is based on large amount of data and computational in nature, rather than formula/model driven. Data Science has currently many different connotations. On one side it praises the "artistry", the genius of laying out connections between disciplines and concepts. This is a truly great aspect of scientists and creativity is definitely very welcome in all data science profiles. With the fun of creating new insights and new data golden eggs, a data scientist has to put up with those annoying criteria of reproducibility, falsifiability and peer reviewing. Sometimes these elements are postponed or left behind in name of the artistry. Granted, it's just hard to find metrics and baselines in order to compare models and data science solutions. But the scientific method has proven to be solid over the centuries and has proven to allow factual scientific discussion between scientists and a to allow selection between models based on objective agreed criteria.

Big and fast a quest for relevant and real-time analytics

Natalino Busa

Big Data and APIs - a recon tour on how to successfully do Big Data analytics

Natalino Busa

This document discusses big data and APIs. It provides examples of the large amount of data generated daily by Facebook and WhatsApp. It then discusses using MapReduce to parallelize processing of big data across distributed systems. It describes challenges like dealing with large sparse matrices and context in big data. It proposes using machine learning and time series analysis to help process big data. Finally, it discusses distributed computing paradigms like Hadoop and architectures used to manage complex, distributed big data systems at large scale.

Strata 2014: Data science and big data trending topics

Natalino Busa

Streaming computing: architectures, and tchnologies

Natalino Busa

Big data landscape

Natalino Busa

An overview about several technologies which contribute to the landscape of Big Data. An intro about the technology challenges of Big Data, follow by key open-source components which help out in dealing with various big data aspects such as OLAP, Real-Time Online Analytics, Machine Learning on Map-Reduce. I conclude with an enumeration of the key areas where those technologies are most likely unleashing new opportunity for various businesses.

More from Natalino Busa (17)

Data Production Pipelines: Legacy, practices, and innovation

[Ai in finance] AI in regulatory compliance, risk management, and auditing

Strata London 16: sightseeing, venues, and friends

Data in Action

Real-Time Anomaly Detection with Spark MLlib, Akka and Cassandra

The evolution of data analytics

Towards Real-Time banking API's: Introducing Coral, a web api for realtime st...

Streaming Api Design with Akka, Scala and Spray

Hadoop + Cassandra: Fast queries on data lakes, and wikipedia search tutorial.

Big data solutions for advanced marketing analytics

Awesome Banking API's

Yo. big data. understanding data science in the era of big data.

Big and fast a quest for relevant and real-time analytics

Big Data and APIs - a recon tour on how to successfully do Big Data analytics

Strata 2014: Data science and big data trending topics

Streaming computing: architectures, and tchnologies

Big data landscape

Recently uploaded

一比一原版(UO毕业证)渥太华大学毕业证如何办理

bmucuha

原件一模一样【微信：95270640】【渥太华大学毕业证UO学位证成绩单】【微信：95270640】（留信学历认证永久存档查询）采用学校原版纸张、特殊工艺完全按照原版一比一制作（包括：隐形水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠，文字图案浮雕，激光镭射，紫外荧光，温感，复印防伪）行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备，十五年致力于帮助留学生解决难题，业务范围有加拿大、英国、澳洲、韩国、美国、新加坡，新西兰等学历材料，包您满意。【业务选择办理准则】一、工作未确定，回国需先给父母、亲戚朋友看下文凭的情况，办理一份就读学校的毕业证【微信：95270640】文凭即可二、回国进私企、外企、自己做生意的情况，这些单位是不查询毕业证真伪的，而且国内没有渠道去查询国外文凭的真假，也不需要提供真实教育部认证。鉴于此，办理一份毕业证【微信：95270640】即可三、进国企，银行，事业单位，考公务员等等，这些单位是必需要提供真实教育部认证的，办理教育部认证所需资料众多且烦琐，所有材料您都必须提供原件，我们凭借丰富的经验，快捷的绿色通道帮您快速整合材料，让您少走弯路。留信网认证的作用: 1:该专业认证可证明留学生真实身份【微信：95270640】 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才 → 【关于价格问题（保证一手价格）我们所定的价格是非常合理的，而且我们现在做得单子大多数都是代理和回头客户介绍的所以一般现在有新的单子我给客户的都是第一手的代理价格，因为我想坦诚对待大家不想跟大家在价格方面浪费时间对于老客户或者被老客户介绍过来的朋友，我们都会适当给一些优惠。选择实体注册公司办理，更放心，更安全！我们的承诺：可来公司面谈，可签订合同，会陪同客户一起到教育部认证窗口递交认证材料，客户在教育部官方认证查询网站查询到认证通过结果后付款，不成功不收费！办理渥太华大学毕业证毕业证offerUO学位证【微信：95270640 】外观非常精致，由特殊纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理渥太华大学毕业证UO学位证毕业证offer【微信：95270640 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理渥太华大学毕业证毕业证offerUO学位证【微信：95270640 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理渥太华大学毕业证毕业证offerUO学位证【微信：95270640 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

Module 1 ppt BIG DATA ANALYTICS_NOTES FOR MCA

yuvarajkumar334

一比一原版马来西亚博特拉大学毕业证（upm毕业证）如何办理

eudsoh

原版一模一样【微信：741003700 】【马来西亚博特拉大学毕业证（upm毕业证）成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理马来西亚博特拉大学毕业证（upm毕业证）【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理马来西亚博特拉大学毕业证（upm毕业证）【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理马来西亚博特拉大学毕业证（upm毕业证）【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理马来西亚博特拉大学毕业证（upm毕业证）【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

一比一原版悉尼大学毕业证如何办理

keesa2

原版一模一样【微信：741003700 】【悉尼大学毕业证成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理悉尼大学毕业证【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理悉尼大学毕业证【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理悉尼大学毕业证【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理悉尼大学毕业证【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

一比一原版(uom毕业证书)曼彻斯特大学毕业证如何办理

osoyvvf

原版一模一样【微信：741003700 】【(uom毕业证书)曼彻斯特大学毕业证成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理(uom毕业证书)曼彻斯特大学毕业证【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理(uom毕业证书)曼彻斯特大学毕业证【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理(uom毕业证书)曼彻斯特大学毕业证【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理(uom毕业证书)曼彻斯特大学毕业证【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

Build applications with generative AI on Google Cloud

Márton Kodok

We will explore Vertex AI - Model Garden powered experiences, we are going to learn more about the integration of these generative AI APIs. We are going to see in action what the Gemini family of generative models are for developers to build and deploy AI-driven applications. Vertex AI includes a suite of foundation models, these are referred to as the PaLM and Gemini family of generative ai models, and they come in different versions. We are going to cover how to use via API to: - execute prompts in text and chat - cover multimodal use cases with image prompts. - finetune and distill to improve knowledge domains - run function calls with foundation models to optimize them for specific tasks. At the end of the session, developers will understand how to innovate with generative AI and develop apps using the generative ai industry trends.

一比一原版卡尔加里大学毕业证（uc毕业证）如何办理

oaxefes

原版一模一样【微信：741003700 】【卡尔加里大学毕业证（uc毕业证）成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理卡尔加里大学毕业证（uc毕业证）【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理卡尔加里大学毕业证（uc毕业证）【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理卡尔加里大学毕业证（uc毕业证）【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理卡尔加里大学毕业证（uc毕业证）【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

社内勉強会資料_Hallucination of LLMs　　　　　　　　　　　　　　　.

NABLAS株式会社

一比一原版(uob毕业证书)伯明翰大学毕业证如何办理

9gr6pty

原版一模一样【微信：6496090 】【(uob毕业证书)伯明翰大学毕业证成绩单】【微信：6496090 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微6496090 【主营项目】一.毕业证【q微6496090】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微6496090】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理(uob毕业证书)伯明翰大学毕业证【微信：6496090 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理(uob毕业证书)伯明翰大学毕业证【微信：6496090 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理(uob毕业证书)伯明翰大学毕业证【微信：6496090 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理(uob毕业证书)伯明翰大学毕业证【微信：6496090 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

[VCOSA] Monthly Report - Cotton & Yarn Statistics May 2024

Vietnam Cotton & Spinning Association

We are pleased to share with you the latest VCOSA statistical report on the cotton and yarn industry for the month of May 2024. Starting from January 2024, the full weekly and monthly reports will only be available for free to VCOSA members. To access the complete weekly report with figures, charts, and detailed analysis of the cotton fiber market in the past week, interested parties are kindly requested to contact VCOSA to subscribe to the newsletter.

[VCOSA] Monthly Report - Cotton & Yarn Statistics March 2024

Vietnam Cotton & Spinning Association

We are pleased to share with you the latest VCOSA statistical report on the cotton and yarn industry for the month of March 2024. Starting from January 2024, the full weekly and monthly reports will only be available for free to VCOSA members. To access the complete weekly report with figures, charts, and detailed analysis of the cotton fiber market in the past week, interested parties are kindly requested to contact VCOSA to subscribe to the newsletter.

一比一原版澳洲西澳大学毕业证（uwa毕业证书）如何办理

aguty

原版一模一样【微信：741003700 】【澳洲西澳大学毕业证（uwa毕业证书）成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理澳洲西澳大学毕业证（uwa毕业证书）【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理澳洲西澳大学毕业证（uwa毕业证书）【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理澳洲西澳大学毕业证（uwa毕业证书）【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理澳洲西澳大学毕业证（uwa毕业证书）【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

Open Source Contributions to Postgres: The Basics POSETTE 2024

ElizabethGarrettChri

Postgres is the most advanced open-source database in the world and it's supported by a community, not a single company. So how does this work? How does code actually get into Postgres? I recently had a patch submitted and committed and I want to share what I learned in that process. I’ll give you an overview of Postgres versions and how the underlying project codebase functions. I’ll also show you the process for submitting a patch and getting that tested and committed.

A gentle exploration of Retrieval Augmented Generation

dataschool1

Digital Marketing Performance Marketing Sample .pdf

Vineet

ML-PPT-UNIT-2 Generative Classifiers Discriminative Classifiers

MastanaihnaiduYasam

Generative Classifiers: Classifying with Bayesian decision theory, Bayes’ rule, Naïve Bayes classifier. Discriminative Classifiers: Logistic Regression, Decision Trees: Training and Visualizing a Decision Tree, Making Predictions, Estimating Class Probabilities, The CART Training Algorithm, Attribute selection measures- Gini impurity; Entropy, Regularization Hyperparameters, Regression Trees, Linear Support vector machines.

Cell The Unit of Life for NEET Multiple Choice Questions.docx

vasanthatpuram

一比一原版(曼大毕业证书)曼尼托巴大学毕业证如何办理

ytypuem

原版一模一样【微信：741003700 】【(曼大毕业证书)曼尼托巴大学毕业证成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理(曼大毕业证书)曼尼托巴大学毕业证【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理(曼大毕业证书)曼尼托巴大学毕业证【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理(曼大毕业证书)曼尼托巴大学毕业证【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理(曼大毕业证书)曼尼托巴大学毕业证【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样

ihavuls

学校原件一模一样【微信：741003700 】《(unimelb毕业证书)墨尔本大学毕业证》【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才

一比一原版斯威本理工大学毕业证（swinburne毕业证）如何办理

actyx

原版一模一样【微信：741003700 】【斯威本理工大学毕业证（swinburne毕业证）成绩单】【微信：741003700 】学位证，留信认证（真实可查，永久存档）原件一模一样纸张工艺/offer、雅思、外壳等材料/诚信可靠,可直接看成品样本，帮您解决无法毕业带来的各种难题！外壳，原版制作，诚信可靠，可直接看成品样本。行业标杆！精益求精，诚心合作，真诚制作！多年品质 ,按需精细制作，24小时接单,全套进口原装设备。十五年致力于帮助留学生解决难题，包您满意。本公司拥有海外各大学样板无数，能完美还原。 1:1完美还原海外各大学毕业材料上的工艺：水印，阴影底纹，钢印LOGO烫金烫银，LOGO烫金烫银复合重叠。文字图案浮雕、激光镭射、紫外荧光、温感、复印防伪等防伪工艺。材料咨询办理、认证咨询办理请加学历顾问Q/微741003700 【主营项目】一.毕业证【q微741003700】成绩单、使馆认证、教育部认证、雅思托福成绩单、学生卡等！二.真实使馆公证(即留学回国人员证明,不成功不收费) 三.真实教育部学历学位认证（教育部存档！教育部留服网站永久可查）四.办理各国各大学文凭(一对一专业服务,可全程监控跟踪进度) 如果您处于以下几种情况： ◇在校期间，因各种原因未能顺利毕业……拿不到官方毕业证【q/微741003700】 ◇面对父母的压力，希望尽快拿到； ◇不清楚认证流程以及材料该如何准备； ◇回国时间很长，忘记办理； ◇回国马上就要找工作，办给用人单位看； ◇企事业单位必须要求办理的 ◇需要报考公务员、购买免税车、落转户口 ◇申请留学生创业基金留信网认证的作用: 1:该专业认证可证明留学生真实身份 2:同时对留学生所学专业登记给予评定 3:国家专业人才认证中心颁发入库证书 4:这个认证书并且可以归档倒地方 5:凡事获得留信网入网的信息将会逐步更新到个人身份内，将在公安局网内查询个人身份证信息后，同步读取人才网入库信息 6:个人职称评审加20分 7:个人信誉贷款加10分 8:在国家人才网主办的国家网络招聘大会中纳入资料，供国家高端企业选择人才办理斯威本理工大学毕业证（swinburne毕业证）【微信：741003700 】外观非常简单，由纸质材料制成，上面印有校徽、校名、毕业生姓名、专业等信息。办理斯威本理工大学毕业证（swinburne毕业证）【微信：741003700 】格式相对统一，各专业都有相应的模板。通常包括以下部分：校徽：象征着学校的荣誉和传承。校名:学校英文全称授予学位：本部分将注明获得的具体学位名称。毕业生姓名：这是最重要的信息之一，标志着该证书是由特定人员获得的。颁发日期：这是毕业正式生效的时间，也代表着毕业生学业的结束。其他信息：根据不同的专业和学位，可能会有一些特定的信息或章节。办理斯威本理工大学毕业证（swinburne毕业证）【微信：741003700 】价值很高，需要妥善保管。一般来说，应放置在安全、干燥、防潮的地方，避免长时间暴露在阳光下。如需使用，最好使用复印件而不是原件，以免丢失。综上所述，办理斯威本理工大学毕业证（swinburne毕业证）【微信：741003700 】是证明身份和学历的高价值文件。外观简单庄重，格式统一，包括重要的个人信息和发布日期。对持有人来说，妥善保管是非常重要的。

Recently uploaded (20)

一比一原版(UO毕业证)渥太华大学毕业证如何办理

Module 1 ppt BIG DATA ANALYTICS_NOTES FOR MCA

一比一原版马来西亚博特拉大学毕业证（upm毕业证）如何办理

一比一原版悉尼大学毕业证如何办理

一比一原版(uom毕业证书)曼彻斯特大学毕业证如何办理

Build applications with generative AI on Google Cloud

一比一原版卡尔加里大学毕业证（uc毕业证）如何办理

社内勉強会資料_Hallucination of LLMs　　　　　　　　　　　　　　　.

一比一原版(uob毕业证书)伯明翰大学毕业证如何办理

[VCOSA] Monthly Report - Cotton & Yarn Statistics May 2024

[VCOSA] Monthly Report - Cotton & Yarn Statistics March 2024

一比一原版澳洲西澳大学毕业证（uwa毕业证书）如何办理

Open Source Contributions to Postgres: The Basics POSETTE 2024

A gentle exploration of Retrieval Augmented Generation

Digital Marketing Performance Marketing Sample .pdf

ML-PPT-UNIT-2 Generative Classifiers Discriminative Classifiers

Cell The Unit of Life for NEET Multiple Choice Questions.docx

一比一原版(曼大毕业证书)曼尼托巴大学毕业证如何办理

原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样

一比一原版斯威本理工大学毕业证（swinburne毕业证）如何办理

7 steps for highly effective deep neural networks

1. 7 steps for highly effective deep neural networks Natalino Busa - Head of Data Science

2. 2 Natalino Busa - @natbusa O’Reilly Author and Speaker Teradata EMEA Practice Lead on Open Source Technologies Teradata Principal Data Scientist ING Group Enterprise Architect: Cybersecurity, Marketing, Fintech Cognitive Finance Group Advisory Board Member Philips Senior Researcher, Data Architect Linkedin and Twitter: @natbusa

3. 3 Natalino Busa - @natbusa X: independent variable Y: dependent variable Linear Regression: How to best fit a line to some data

4. 4 Natalino Busa - @natbusa input : 784 numbers output: 10 classes 28x28 pixels Classification: Handwritten digits

5. 5 Natalino Busa - @natbusa Sharing the (Not so) Secret Lore: Keras & Tensorflow Some smaller projects: Tflearn, Tensorlayers http://keras.io/

6. 6 Natalino Busa - @natbusa 1: Single Layer Perceptron “dendrites” Axon’s response Activation function

7. 7 Natalino Busa - @natbusa More activation functions: https://en.wikipedia.org/wiki/Activation_function

8. 8 Natalino Busa - @natbusa Single Layer Neural Network Takes: n-input features: Map them to a soft “binary” space ∑ x1 x2 xn f 1: Single Layer Perceptron (binary classifier)

9. 9 Natalino Busa - @natbusa 1: Single Layer Perceptron (multi-class classifier) Values between 0 and 1 Sum of all outcomes = 1 It produces an estimate of a probability! From soft binary space to predicting probabilities: Take n inputs, Divide by the sum of the predicted values ∑ x1 x2 xn f ∑ f softmax ‘1’: 95% ‘8’: 5%

10. 10 Natalino Busa - @natbusa 1: Single Layer Perceptron Minimize costs: The cost function depends on: - Parameters of the model - How the model “composes” Goal : Reduce the mean probability error modify the parameters to reduce the error! Vintage math from last century

11. 11 Natalino Busa - @natbusa Supervised Learning Stack layers of perceptrons - Feed Forward Network - Scoring goes from input to output - Back propagate the error from output to input SOFTMAX Input parameters classes (estimated probabilities) Feed-forwardfunctions Cost function supervised : actual output BackPropagateErrors

12. 12 Natalino Busa - @natbusa Let’s go!

13. 13 Natalino Busa - @natbusa 1. Single Layer Perceptron

14. 14 Natalino Busa - @natbusa 1. Single Layer Perceptron

15. 15 Natalino Busa - @natbusa 1. Single Layer Perceptron

16. 16 Natalino Busa - @natbusa 1. Single Layer Perceptron

17. 17 Natalino Busa - @natbusa Tensorboard

18. 18 Natalino Busa - @natbusa Tensorboard

19. 19 Natalino Busa - @natbusa Tensorboard

20. 20 Natalino Busa - @natbusa 2. Multi Layer Perceptron

21. 21 Natalino Busa - @natbusa 2. Multi Layer Perceptron

22. 22 Natalino Busa - @natbusa 2. Multi Layer Perceptron

23. 23 Natalino Busa - @natbusa 2. Multi Layer Perceptron

24. 24 Natalino Busa - @natbusa 3. Convolution From Krizehvsky et al. (2012)

25. 25 Natalino Busa - @natbusa 3. Convolution diagrams: By Aphex34 - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=45659236 By Aphex34 - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=45673581 CC0, https://en.wikipedia.org/w/index.php?curid=48817276 convolution Max pooling RELU / ELU

26. 26 Natalino Busa - @natbusa 3. Convolution

27. 27 Natalino Busa - @natbusa 3. Convolution

28. 28 Natalino Busa - @natbusa 3. Batch Normalization

29. 29 Natalino Busa - @natbusa 3. Batch Normalization ( example for MLP)

30. 30 Natalino Busa - @natbusa 3. Batch Normalization ( example for MLP)

31. 31 Natalino Busa - @natbusa 3. Batch Normalization ( example for MLP) Activation function: SIGMOID

32. 32 Natalino Busa - @natbusa 3. Batch Normalization ( example for MLP) Activation function: RELU

33. 33 Natalino Busa - @natbusa 4. Regularization: Prevent overfitting in ANNs - Batch Normalization - RELU/ELU - RESIDUAL / SKIP Networks - DROP LAYER - REDUCE PRECISION (HUFFMAN ENCODING) In general ANN are parameters rich, constraining the parameter space usually produces better results and speed up the learning

34. 34 Natalino Busa - @natbusa 5. Inception architectures Cannot be stacked!

35. 35 Natalino Busa - @natbusa 5. Inception architectures

36. 36 Natalino Busa - @natbusa 5. Inception architectures compression Avoid dimensions explosion

37. 37 Natalino Busa - @natbusa 5. Inception architectures

38. 38 Natalino Busa - @natbusa 5. Inception architectures (top architecture)

39. 39 Natalino Busa - @natbusa 6. Residual Networks https://culurciello.github.io/tech/2016/06/04/nets.html

40. 40 Natalino Busa - @natbusa 6. Residual Networks

41. 41 Natalino Busa - @natbusa 6. Residual + Inception Networks

42. 42 Natalino Busa - @natbusa 7. LSTM on Images

43. 43 Natalino Busa - @natbusa 7. LSTM on Images

44. 44 Natalino Busa - @natbusa 7. LSTM on Images http://colah.github.io/posts/2015-08-Understanding-LSTMs/

45. 45 Natalino Busa - @natbusa 7. LSTM on Images

46. 46 Natalino Busa - @natbusa 7. LSTM on Images

47. 47 Natalino Busa - @natbusa 7. LSTM on ConvNets (bonus slide)

48. 48 Natalino Busa - @natbusa All the codez :) https://github.com/natbusa/deepnumbers

49. 49 Natalino Busa - @natbusa Meta- References … just a few articles, but extremely dense in content. A must read! https://keras.io/ http://karpathy.github.io/neuralnets/ https://culurciello.github.io/tech/2016/06/04/nets.html http://colah.github.io/posts/2015-08-Understanding-LSTMs/ https://gab41.lab41.org/batch-normalization-what-the-hey-d480039a9e3b

7 steps for highly effective deep neural networks

Recommended

Recommended

More Related Content

Similar to 7 steps for highly effective deep neural networks

Similar to 7 steps for highly effective deep neural networks (20)

More from Natalino Busa

More from Natalino Busa (17)

Recently uploaded

Recently uploaded (20)

7 steps for highly effective deep neural networks