Intending to read the entire Foundation universe, I recently started with the robots series. Finished I, Robot and Caves of Steel, and halfway through The Naked Sun. I've read the latter a long time ago as a standalone, so I know the plot.
While reading these, one common theme that pops out is that Asimov's robots - spanning from now ish to few thousand years from now - seem to have made no progress in, and are terrible at, nature language processing. And this fact seems to be a recurring major plot point/devise as well.
I'm wondering why this is. Our current tech such as Google Assistant, Alexa, and Siri have come a long way in this regard compared to a decade ago, although they're far from perfect. I understand that there's a 70 year gap between then and now, and a lot has changed beyond imagination particularly when it comes to computers. But still, as a Sci-Fi writer couldn't he have imagined that machines would get good at it over the course of 3,000 years? Was this the common thought in the 50s, or was it just Asimov? Given that Clarke's 2001 has a very human like AI, and it's written in early 60s, I guess it can't be so hard to imagine human like robots?
I haven't read the Empire or Foundation series, so appreciate no spoilers in comments.
Hey!
I am looking for a Suicide detection dataset.. Anybidy has any idea, on where can I get that ?
I need it to perform experimentations for my college disseration, as I have always worked on NLP. If anyone has any idea, where I can find one it'd be helpful. Also, I can't use API and then have it annotated, as I tried before but they won't accept it.
Either for studying or for outside class
GPT-3 is the advanced natural language processing model developed by OpenAI. It returns a natural language text completion in response to any text request, such as a phrase or a sentence. Developers use GPT-3 (through on-demand charging via application programming (API)) in their applications to do tasks such as text translation and software code development.
OpenAI has recently released new functionality that will allow developers to create their own versions of GPT-3. The new customization option is now available in the API.
GPT-3 can execute a wide range of natural language tasks with just a few instances, a notion known as few-shot learning or prompt design. GPT-3 can be customized to produce much better results because it allows users to provide far more instances than prompt design allows.
Get Access: https://beta.openai.com/docs/guides/fine-tuning/preparing-your-dataset
Open AI Blog: https://openai.com/blog/customized-gpt3/
We are seeking a Research Scientist to work on Natural Language Processing (NLP) projects that leverage state-of-the-art deep learning models, high performance computing, and large datasets of unstructured clinical text. Example problems include extracting topography and morphology information from cancer pathology reports, using the Summit supercomputer to pretrain state-of-the-art Transformer language models for clinical and biomedical text, and developing privacy-preserving deep learning models that generalize well to clinical organizations across the US. There is flexibility in defining new research directions relevant to the problems and projects being tackled. The position will be in the Biostatistics and Multiscale Systems (BMS) group in the Advanced Computing in Health Science (ACHS) Section of the Computational Science and Engineering Division (CSED).
As a research scientist, you will have the opportunity to help solve some of the most challenging problems this world faces. You will perform ground breaking research on a wide range of significant problems with the fastest computing platforms in the world. This position requires novel thinking, teamwork, and discovery in finding new approaches for analyzing massive and complex data, collaborating with worldwide experts, and publishing groundbreaking results.
Read more / apply: https://ai-jobs.net/job/12152-research-scientist-in-biomedical-natural-language-processing-and-deep-learning/
I've heard it's relatively easy for undergrads who are majoring cs in cmu to participate in ML/CV/NLP researches. But is it true for cs masters students as well? Since cmu is literally the best when it comes to ML/AI research, I think there will be a lot of opportunities for masters students as well.
I’m Sourabh, I lead one of the core Tensorflow teams at Google Brain and worked on data products at Coursera with Andrew Ng. Kaushik Rangadurai, ML Engineer at Facebook and I are leading a live, cohort based course on NLP starting November 1st. https://corise.com/course/natural-language-processing.
We wanted to share what we’ve learned in machine learning over the years. You can join the first run of the course (capped at about 30 students) below. If you’re open to giving feedback on the class on how we can do better, happy to give a discount.
Hi. There is something lately I am searching about. Google uses anchor texts and its neighboring texts to rank specific pages. Nowadays, it seems google uses natural language processing for this purpose. Can anybody who searched on this task before helping introduce and show me some resources in this field.
