Writing Code for NLP Research

Slides by the Allen Institute for Artificial Intelligence: https://lnkd.in/eubgSGY

#naturallanguageprocessing #NLP #research

If you like our channel, i invite you to share it with your friends

🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Best Paper Awards in Computer Science (since 1996)

A well maintained list: https://lnkd.in/e6_ks3E

#artificialintelligence #machinelearning #papers #research

🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

A new study published in Nature shows we can predicting the genetic disorder directly from the face using deep learning. The network was trained on a dataset of 17,000 patient images representing more than 200 syndromes. The paper reports that model achieves 91% top-10-accuracy in identifying the correct syndrome on 502 images and outperformed expert clinicians in three experiments. The method can be used to diagnose dysmorphology syndromes which typically affect roughly 1 in 30,000 people.
While this work has a great potential to improve discovering rare diseases, insurance companies may use this technology to deny providing medical insurance or increase the policy fees for people with specific genes.
paper: https://lnkd.in/fUEpYRt
#artificialintelligence #syndrome #ai #deeplearning #research

🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Top 10 #deeplearning research papers as per this website
https://lnkd.in/dPYayt9

Of course the choice remains biased but we do like these besides a few hundred other papers.

Remember, it is not the popular but the meaningful and industry relevant research that is worth paying attention to.

Here's the list:

1. Universal Language Model Fine-tuning for Text Classification
https://lnkd.in/dhj5SyM

2. Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples
https://lnkd.in/d44kt3Q

3. Deep Contextualized Word Representations
https://lnkd.in/dkP68Fb

4. An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
https://lnkd.in/dAhYzge

5. Delayed Impact of Fair Machine Learning
https://lnkd.in/dvTvG2s

6. World Models

7. Taskonomy: Disentangling Task Transfer Learning
https://lnkd.in/dYxMjAd

8. Know What You Don’t Know: Unanswerable Questions for SQuAD
https://lnkd.in/d--grME

9. Large Scale GAN Training for High Fidelity Natural Image Synthesis
https://lnkd.in/dY6psf4

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
https://lnkd.in/dgtnD7n
#machinelearning #research #deeplearning #artificialintelligence

🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

The Evolved Transformer

Paper by So et al.: https://lnkd.in/eNZ6ije

#artificalintelligence #MachineLearning #NeuralComputing #EvolutionaryComputing #research

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Our friends in Tampere University are hiring now:
15 fully-funded PhD job openings for 3 years within H2020 A-WEAR European Joint Doctorate

Target audience: fresh MSc graduates in various engineering fields (who have completed their first master no earlier than Fall 2015) and who are passionate about pursuing a PhD in a research field of high relevance to today’s society (wearable computing & IoT).


Job description: fully funded 36 months PhD positions towards double/joint PhD programs in 5 top European technical universities in Finland, Italy, Spain, Czech Republic, and Romania

Gross salary (approx. in EUR/month): 3600 (FI), 2800 (ES), 2000 (RO), 2400 (CZ), 2900 (IT)

Application deadline: 28th of February 2019

Starting time of the PhD: Fall 2019
https://www.tuni.fi/en
https://lnkd.in/eyDattx

#universities #graduations #phd #funding #research

✴️ @AI_Python_EN
❇️ @AI_Python
🗣 @AI_Python_arXiv

TensorFlow 2.0 is the best bet for Deep Learning Community.

Eager execution for easy prototyping & debugging along with tf.function() advantage,

Distribution Strategies for distributed Training (including multi node, multi accelerator including TPU pods, also Kubernetes),

Smoother building, training,validation with tf.keras and premade Estimators,

Smart deployment (TensorFlow Serving(A TensorFlow library allowing models to be served over HTTP/REST), TensorFlow Lite(TensorFlow’s lightweight solution for mobile and embedded devices), TensorFlow.js(Enables deploying models in JavaScript environments, such as in a web browser or server side through Node.js), TensorFlow Hub),
Compatiable with TF 1.x (also a conversion tool which updates TensorFlow 1.x Python code to use TensorFlow 2.0 compatible APIs, or flags cases where code cannot be converted automatically )

Also great for researchers ( Model Subclassing API, automatic differentiation, Ragged Tensors, TensorFlow Probability, Tensor2Tensor)

For beginners, TensorFlow, https://lnkd.in/fp3AWKk

#tensorflow #research #deeplearning #pyTorch

✴️ @AI_Python_EN

OpenAI has created activation atlases (in collaboration with Google researchers), a new technique for visualizing what interactions between neurons can represent.

As AI systems are deployed in increasingly sensitive contexts, having a better understanding of their internal decision-making processes will let us identify weaknesses and investigate failures.

Blog: https://lnkd.in/d4i6xQC
Paper: https://lnkd.in/dGNcd4K
Github: https://lnkd.in/d-2WhfN
Demo: https://lnkd.in/dBiHZv3

#deeplearning #research

✴️ @AI_Python_EN

Do you remember a bullshit study published a few years ago claiming that deep learning can spot criminals from their photos and arguing that criminals have different facial features. Despite the ethical issue, we know this is bullshit but we couldn't spot the flaws.
Well, like most machine learning problems the devil is in the data.
To train the model the researchers used 700 of criminals ID photos as positive images. On other hands, they collected 1100 non-criminals from the web which featured people smiling.
No wonder why they go 90% accuracy!
So instead of developing criminals detector, they developed smiles detector LOL.
#research #machinelearning #deeplearning #ai
https://lnkd.in/fMhU4ZZ

✴️ @AI_Python_EN

Hacking Google reCAPTCHA v3 using Reinforcement Learning

Paper: https://lnkd.in/es9AjzC

#reinforcementlearning #research #ai #artificialintelligence #machinelearning

✴️ @AI_Python_EN