dyad x machina

a tiny lab at the edges of affective neuroscience x deep learning 

 
pickles-editorial-green-v6 (2).jpg

dyad x machina

Emotions represent complex neural processes that lie at the helm of humanity's best and worst. We study affective neuroscience, do basic research in psychophysiology, and use deep learning with the goal of creating "the affective layer"  which we believe will fuel tomorrow's technologies. 

The affective layer is what we use to guide ourselves in the world. When we make a decision, we not only analyse the potential pros and cons using cognition (think prefrontal cortex), we also feel through possible futures to come to a final decision (think limbic system to VMPFC). We all know how friends can influence our purchases, but what isn't as obvious is the complex array of negative and positive emotions that move us to act and finally press the "buy" button.

MOVING FROM LAB > FIELD > MACHINE LEARNING MODEL

Tomorrow's applications should adapt to the user's affective state in real-time rather than act as if the user is a monolithic persona that rarely changes.

This is typically quantified in the lab using measures like EDA, EEG, HRV, and fMRI technology (or even transcranial magnetic stimulation), but it can also be quantified out in the field and in applications through physiological wristbands like the E4 or a fitbit.

Deep learning and TensorFlow has allowed us to use computer vision (i.e. faces), audio, and complex physiological signal processing to better classify affective states and model their neural substrates.  

Research has also been done using clickstream data and in-app feature usage to create "affective" features to feed into models without the need for physiological sensors, we are interested in this work as well.

FROM LAB > FIELD > MACHINE LEARNING MODEL

We'd love to collaborate to help bring the affective layer to life. 

 

"Our goal is to design the affective layer into technology for people and groups"

— dyadxmachina

Research

In 2014, we dove deeply into the emotion sciences, consuming 1000's of papers and foundational books (like this one and this one) from cognitive science to affective neuroscience to understand why we had such a thing as fickle as emotions and why they were so important.

We currently run other animal and human experiments to understand complex human emotions like shame and pride. and create machine learning systems that will allow us to begin to continually surface our internal affective processes with applications in biofeedback, emerging tech (vr, ar), mental health, and fuel modern applications.  We are wildly curious, seriously disciplined, and dedicated to learning more about the "affective" layer of our lives.

We work with partners to understand emotion x ___ by running experiments that help us better understand the affective layer.

Applications

We are interested in taking the best of affective neuroscience and deep learning to  inform the design of applications that take a user's emotion (biomarkers, clickstream, image, audio, video) as an input into modern apps as well as an output for biofeedback. 

  TensorFlow   To process noisy and high dimensional physiological data, we use TensorFlow to train, evaluate, and deploy our deep learning systems.  Checkout our course on  Applied Deep Learning with TensorFlow and Google Cloud AI  

TensorFlow

To process noisy and high dimensional physiological data, we use TensorFlow to train, evaluate, and deploy our deep learning systems.

Checkout our course on Applied Deep Learning with TensorFlow and Google Cloud AI 

  Empatica    We use Empatica's E4 technology for long term (months) ambulatory monitoring of psychophysiological signals to measure stress and affect over time.  We'll be developing a course on Affective Computing and TensorFlow to help developers who want to learn about emotion processing and researchers who want to take their research into production.      

Empatica

We use Empatica's E4 technology for long term (months) ambulatory monitoring of psychophysiological signals to measure stress and affect over time.

We'll be developing a course on Affective Computing and TensorFlow to help developers who want to learn about emotion processing and researchers who want to take their research into production.

 

 

  Biopac Systems   We rely on Biopac for our controlled experiments because of it's high resolution and overall robustness.

Biopac Systems

We rely on Biopac for our controlled experiments because of it's high resolution and overall robustness.

Work with us

FROM LAB > FIELD > MACHINE LEARNING MODEL

We are working with partners to conduct specific research (i.e. decision making and emotion) and designing the affective layer to enable intelligent apps, mental health, and putting emotion back up there with cognition.

 
pickles-editorial-blue-v2.jpg

Contact Us

Please feel free to reach out about our research, insights, or possible partnerships that you think might be interesting for yourself, organisation, or company. 

We are interesting in working on affective computing projects of all sorts but especially in the biofeedback and adaptive application areas.

Email: two@dyadxmachina.com

 

 

 

Located in 

Palo Alto, CA
United States