Categories
Articles

Top Twitter Topics by Data Scientists #38

Trending this week: What is Deep Reasoning? Use Deep Learning to efficiently identify visually similar patterns from pairs of images; Is Deep Learning worthy for time-series forecasting?

Every week we analyze the most discussed topics on Twitter by Data Science & AI influencers.

The following topics, URLs, resources, and tweets have been automatically extracted using a topic modeling technique based on Sentence BERT, which we have enhanced to fit our use case.

Want to know more about the methodology used? Check out this article for more details, and find the codes in this Github repository!

Overview

This post, dedicated to the technology watch based on posts from Data Science and AI influencers on Twitter, will cover various topics this week. In particular, we will talk about:

  • Deep Learning Updates
  • Deep Learning for Times-Series Forecasting
  • Social Media & Digital Marketing

Deep Learning Updates

This week, data science and AI influencers have shared some updates on deep learning.

Yann LeCun has shared a research paper introducing DeepMP, a deep learning tool to detect DNA modifications on Nanopore sequencing data. DeepMP is a convolutional neural network (CNN)-based model that takes information from Nanopore signals and basecalling errors to detect whether a given motif in a read is methylated or not. Besides, DeepMP introduces a threshold-free position modification calling model sensitive to sites methylated at low frequency across cells. DeepMP is freely available on Github.

Tomasz Malisiewicz has posted a research paper on Learning Co-segmentation by Segment Swapping for Retrieval and Discovery. This paper demonstrates how to efficiently identify visually similar patterns from a pair of images, e.g. identifying an artwork detail copied between an engraving and an oil painting, or matching a night-time photograph with its daytime counterpart. The approach presented in this work has demonstrated clear improvements for artwork details retrieval on the Brueghel dataset and achieved competitive performance on two place recognition benchmarks, Tokyo247 and Pitts30K. Also, it has demonstrated high potential for performing object discovery on the Internet object discovery dataset and the Brueghel dataset.

Terence Mills has tweeted about Deep Reasoning. The concept of deep reasoning is said to bring us all one step closer to artificial general intelligence. It is said to be the next level beyond deep learning. It will allow AI models to make decisions based on the theory called common sense i.e. abstract reasoning. It helps machines understand implied relationships between different things. For example, using deep reasoning, an AI will be able to solve or answer questions like — “what is the size of that ball which is placed at the left side of the cupboard down on the floor?”. Which cannot be answered out of a deep learning method currently.

Will deep reasoning provide human reasoning to machines? (Source: Pixabay)

Deep Learning for Time-Series Forecasting

Also, some very interesting resources were shared on applying deep learning for time-series data forecasting.

Kirk Borne has shared the following content:

A post discussing if deep learning is worthy to solve time-series forecasting problems. This article is the first of a two-part series that aims to provide a comprehensive overview of the state-of-art deep learning models that have proven to be successful for time series forecasting. It focuses on RNN-based models and DeepAR, whereas the second explores transformer-based models for time series. Each article compares these models to the standard forecasting approaches.

This post will provide you an understanding of:

  1. What are the key common concepts of deep learning models used for forecasting time series?
  2. How do RNN-based and DeepAR models differ from one another?
  3. To what extent do they provide better results in terms of forecasting accuracy and computing time?

A blog post discussing ARIMA/SARIMA vs LSTM with Ensemble learning Insights for Time Series Data. The blog is structured as follows:

  • Understanding deep learning algorithms RNN, LSTM, and the role of ensemble learning with LSTM to aid in performance improvement.
  • Understanding conventional time series modeling technique ARIMA and how it helps to improve time series forecasting in ensembling methods when used in conjunction with MLP and multiple linear regression.
  • Understanding problems and scenarios where ARIMA can be used vs LSTM and the pros and cons behind adopting one against the other.
  • Understanding how time series modeling with SARIMA can be clubbed with other spatial, decision-based, and event-based models using ensemble learning.

The study concludes with some case studies on why specific machine learning methods perform so poorly in practice, given their impressive performance in other areas of artificial intelligence. The challenge leaves it open to evaluating reasons for poor performance for ARIMA/SARIMA and LSTM models, and devise mechanisms to improve the model’s poor performance and accuracy.

A podcast on applying deep learning techniques for time-series forecasting. This podcast was realized with Dr. Francesca Lazzeri, a Principal Data Scientist Manager at Microsoft, who shared best practices on implementing deep learning-based solutions to solve time-series forecasting problems.

The main takeaways are:

  • Data Science Lifecycle phases and the importance of each phase in the overall process.
  • How to develop and deploy ML models for time-series forecasting.
  • Best practices when developing machine learning applications on time series forecasting
  • Deep learning techniques for time series forecasting

Social Media & Digital Marketing

This week data science and AI influencers tweeted about Digital marketing and its impact on business.

Tamara McCleary shared the following:

Illustration of a workflow strategy in marketing. (Source: Unsplash)